WO2021082488A1 - Text matching-based intelligent interview method and apparatus, and computer device - Google Patents
Text matching-based intelligent interview method and apparatus, and computer device Download PDFInfo
- Publication number
- WO2021082488A1 WO2021082488A1 PCT/CN2020/098796 CN2020098796W WO2021082488A1 WO 2021082488 A1 WO2021082488 A1 WO 2021082488A1 CN 2020098796 W CN2020098796 W CN 2020098796W WO 2021082488 A1 WO2021082488 A1 WO 2021082488A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- text
- interview
- vector
- word
- complex
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 70
- 239000013598 vector Substances 0.000 claims abstract description 301
- 230000009466 transformation Effects 0.000 claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims description 131
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000006243 chemical reaction Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 16
- 239000000203 mixture Substances 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000012545 processing Methods 0.000 abstract description 14
- 230000008569 process Effects 0.000 description 21
- 230000006870 function Effects 0.000 description 9
- 230000008901 benefit Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000011176 pooling Methods 0.000 description 2
- 230000007115 recruitment Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/105—Human resources
- G06Q10/1053—Employment or hiring
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Definitions
- This application relates to the field of artificial intelligence technology, in particular to an intelligent interview method, device and computer equipment based on text matching.
- the main purpose of this application is to provide a natural language time word parsing method, device and computer equipment, aiming to solve the disadvantages of the existing time word parsing method that is too rigid, accurate and low in completeness.
- the main purpose of this application is to provide an intelligent interview method, device and computer equipment based on text matching, aiming to solve the shortcomings of existing interview methods that lack uniformity and objectivity.
- this application provides an intelligent interview method based on text matching, including:
- interview text is a text formed after the interviewer answers to the interview question
- standard text is a text of a standard answer corresponding to the interview question
- this application also provides an intelligent interview device based on text matching, including:
- An acquisition module for acquiring interview text and standard text where the interview text is the text formed by the interviewer after answering the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
- a conversion module configured to perform vector conversion on the interview text and the standard text respectively according to a first preset rule to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
- the matching module is used to match the corresponding interview scores according to the similarity.
- the present application also provides a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned text matching-based intelligent interview method when the computer program is executed, wherein:
- the intelligent interview method based on text matching includes the following steps:
- interview text is a text formed after the interviewer answers to the interview question
- standard text is a text of a standard answer corresponding to the interview question
- the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the above methods are realized.
- a text matching-based intelligent interview method includes the following steps:
- interview text is a text formed after the interviewer answers to the interview question
- standard text is a text of a standard answer corresponding to the interview question
- This application provides an intelligent interview method, device and computer equipment based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into the corresponding complex-valued vector, and then the complex-valued vector is passed through the corresponding
- the calculation of the interview text and the standard text respectively corresponds to the mixed density matrix, and the maximum probability vector is selected from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then the first vector and the second vector are calculated by The cosine between the two vectors obtains the similarity between the interview text and the standard text, and finally the corresponding interview score is obtained according to the similarity matching.
- the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large.
- the range is improved to realize the high accuracy and objectivity of the smart interview.
- FIG. 1 is a schematic diagram of the steps of an intelligent interview method based on text matching in an embodiment of the present application
- Figure 2 is a block diagram of the overall structure of an intelligent interview device based on text matching in an embodiment of the present application
- FIG. 3 is a schematic block diagram of the structure of a computer device according to an embodiment of the present application.
- an embodiment of the present application provides an intelligent interview method based on text matching, including:
- S1 Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
- S2 Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
- the interview system will sequentially output interview questions to the interviewer for answers according to preset settings, and the interviewer can answer the interview questions by voice or manual input.
- the interview system collects the interviewer's answers to the interview question, it forms an interview text, and at the same time calls the pre-entered standard answer corresponding to the interview question, that is, the standard text.
- the interview system needs to convert the interview text and standard text into corresponding vectors in order to compare the similarity between the two later. Among them, the interview system converts the interview text and standard text into corresponding vectors in the same way, and the conversion between the two
- the two can be converted at the same time, or the standard text can be converted first, and then the interview text can be converted.
- the interview system uses one-hot encoding and performs Complex-valued Embedding on each word in the interview text, so that each word generates a corresponding complex-valued embedding. Value vector, each complex value vector is combined to form a word matrix.
- the interview system uses a sliding window for the word matrix, and each time from the word matrix, according to the arrangement order of the respective words corresponding to the complex-valued vector in the interview text, a preset number of complex-valued vectors are successively selected as the first complex-valued vector and formed The first word matrix until the selection of all complex-valued vectors is completed.
- the complex-valued vector in the word matrix is (a, b, c, d, e)
- the preset number is 3
- the first word matrix selected are: (a, b, c), (b, c) ,d),(c,d,e).
- the interview system needs to perform the same data processing on the filtered first word matrix each time.
- the specific processing steps are: the interview system respectively multiplies each first complex-valued vector with each corresponding conjugate transpose vector to obtain the outer product , So that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix.
- the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector.
- the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm.
- the interview system After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors.
- the interview system adds up the weighted word density matrix to obtain the mixed density matrix.
- the interview system calls the third preset formula, and converts the projection length of the mixed density matrix on each different projection plane into the corresponding second probability, thereby obtaining the multiple second probabilities of the mixed density matrix projected on each projection plane. Probability vector.
- x i > tr( ⁇
- the interview system takes out the largest probability vector among the multiple probability vectors in each sliding window to form the first vector. For example, the probability vector of sliding window A is (1,2,3), and the probability vector of sliding window B is (1,2,3).
- the probability vector is (4,5,6), and the probability vector of the sliding window C is (7,8,9). Then select 3, 6, 9 from the sliding windows A, B, and C to form the first vector (3, 6, 9).
- the interview system converts the standard text into the corresponding second vector.
- the interview system calculates the cosine value between the first vector and the second vector based on the angle between the two, and the calculated cosine value is the similarity between the interview text and the standard text.
- the interview system has pre-entered a similarity and interview score mapping relationship table, so the interview system can obtain the interviewer’s evaluation score in this interview question according to the similarity calculated in the current time.
- the step of performing vector transformation on the interview text according to the first preset rule to obtain the first vector corresponding to the interview text includes:
- S21 Perform complex-valued embedding of each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;
- S24 Select the largest probability vector in each of the sliding windows respectively, and each of the largest probability vectors composes the first vector.
- the interview system uses one-hot code encoding and complex-valued embedding each word in the interview text, so that each word generates a corresponding complex-valued vector, and each complex-valued vector is combined to form a word matrix.
- the word vectors represented by complex-valued vectors not only consider ordinary amplitude addition and subtraction, but also consider the higher-order semantics brought by their phase information, which can achieve the effect of adding two words to have more meanings. The addition of two words can also be counterproductive.
- the interview system uses a sliding window for the word matrix, and each time from the word matrix, according to the arrangement order of the respective words corresponding to the complex-valued vector in the interview text, a preset number of complex-valued vectors are successively selected as the first complex-valued vector and formed The first word matrix, until the selection of all complex-valued vectors is completed, for example, the complex-valued vector in the word matrix is (a,b,c,d,e), and the preset number is 3, then select the first word matrix obtained They are: (a, b, c), (b, c, d), (c, d, e).
- the interview system needs to perform the same data processing on the filtered first word matrix each time.
- the specific processing steps are: the interview system respectively multiplies each first complex-valued vector a by the corresponding conjugate transpose vector aT to get The outer product, so that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix.
- the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector.
- the interview system respectively substitutes the norm corresponding to each first complex-valued vector into a second preset formula, and calculates the first probability corresponding to each norm.
- the interview system After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors.
- the interview system adds up the weighted word density matrix to obtain the mixed density matrix.
- the interview system calls the third preset formula, converts the projection length of the mixed density matrix on each different projection plane into the corresponding second probability, thereby obtaining the probability composed of multiple second probabilities of the mixed density matrix projected on each projection plane vector.
- the interview system takes out the largest probability vector among the multiple probability vectors in each sliding window to form the first vector.
- the step of using a sliding window to convert the word matrix into a mixed density matrix includes:
- S223 Calculate and obtain a mixed density matrix according to each of the word density matrixes and each of the first probabilities.
- the interview system uses a sliding window, and each time a preset number of complex-valued vectors are sequentially selected as the first complex-valued vector from the word matrix according to the sequence of the respective words corresponding to the complex-valued vectors in the interview text. And compose the first word matrix until the selection of all the complex-valued vectors is completed.
- the complex-valued vector in the word matrix is (a,b,c,d,e)
- the preset number is 3, then select the first
- the word matrix is: (a, b, c), (b, c, d), (c, d, e).
- the interview system needs to perform the same data processing on the filtered first word matrix each time.
- the specific processing steps are: the interview system respectively multiplies each first complex-valued vector a by the corresponding conjugate transposed vector a T The outer product is obtained, so that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix.
- the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector.
- the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm.
- the interview system After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors.
- the interview system adds up the weighted word density matrix to obtain the mixed density matrix.
- this embodiment performs weighting based on local words, so that the system has different weights for different words, and can make judgments on words in combination with context.
- step of calculating the first probability corresponding to each of the first complex-valued vectors includes:
- S2222 Substitute each of the norms into a second preset formula, and calculate the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector.
- the second preset formula for: ⁇ (w i) is the norm requires determined, x is the value of the first complex-valued vectors.
- the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm, where the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- the step of calculating a mixed density matrix according to the word density matrix and each of the first probabilities includes:
- the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain the respective weighted word density matrices composed of the calculated vectors.
- the interview system adds up the weighted word density matrix to obtain the mixed density moment.
- this embodiment performs weighting based on local words, so that the system has different weights for different words, and can make judgments on words in combination with context.
- the step of calculating according to a preset algorithm to obtain a plurality of probability vectors corresponding to the mixing density matrix in different sliding windows respectively includes:
- the interview system calls the third preset formula, converts the projection length of the mixed density matrix on the projection plane into the corresponding second probability, and the interview system composes the obtained second probabilities to obtain the mixed density matrix in The probability vector of the projection plane corresponding to the current sliding window.
- x i > tr( ⁇
- step of calculating the similarity between the first vector and the second vector includes:
- the first vector of the interview text conversion and the second vector of the standard text conversion are the vectors obtained after the N-grams of the two texts are projected on the same projection plane, so these two vectors can be calculated Foundation.
- This embodiment provides an intelligent interview method based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into a corresponding complex-valued vector, and then the complex-valued vector is calculated correspondingly to obtain the interview text And the standard text respectively corresponding to the mixed density matrix, and select the largest probability vector from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then calculate the difference between the first vector and the second vector Cosine obtains the similarity between the interview text and the standard text, and finally obtains the corresponding interview score according to the similarity matching.
- the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large.
- the range is improved to realize the high accuracy and objectivity of the smart interview.
- an embodiment of the present application also provides an intelligent interview device based on text matching, including:
- the obtaining module 1 is used to obtain interview text and standard text, where the interview text is the text formed by the interviewer after answering the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
- the conversion module 2 is configured to perform vector conversion on the interview text and the standard text respectively according to a first preset rule to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
- the calculation module 3 is used to calculate the similarity between the first vector and the second vector
- the matching module 4 is used to match the corresponding interview scores according to the similarity.
- the realization process of the functions and functions of the acquisition module 1, the conversion module 2, the calculation module 3 and the matching module 4 in the above-mentioned smart interview device is detailed in the corresponding steps S1 to S4 in the above-mentioned text matching-based smart interview method.
- the realization process of, I will not repeat it here.
- the conversion module 2 includes:
- the embedding sub-module is used to perform complex-valued embedding of each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;
- the first calculation sub-module is configured to calculate according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;
- the selection sub-module is configured to select the largest probability vector in each of the sliding windows, and each of the largest probability vectors composes the first vector.
- the implementation process of the functions and roles of the embedded sub-module, the conversion sub-module, the first calculation sub-module, and the selection sub-module in the above-mentioned smart interview device is detailed in the corresponding step S21 in the above-mentioned text-based smart interview method.
- the implementation process to S24 will not be repeated here.
- transformation sub-module includes:
- the selection unit is configured to use the sliding window to sequentially select a preset number of first complexes from the word matrix according to the order in which the words corresponding to each of the complex-valued vectors are arranged in the interview text.
- the value vectors form a matrix until the selection of all the complex value vectors is completed, and a number of first word matrices are obtained;
- the first calculation unit is configured to calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, and convert each of the first word matrices into corresponding A word density matrix, and calculating the first probability corresponding to each of the first complex-valued vectors;
- the second calculation unit is configured to calculate a mixture density matrix according to each of the word density matrix and each of the first probabilities.
- the implementation process of the functions and roles of the selection unit, the first calculation unit and the second calculation unit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S221 to S223 in the above-mentioned text matching-based intelligent interview method. , I won’t repeat it here.
- the first calculation unit includes:
- the first calculation subunit is configured to substitute each of the first complex-valued vectors into a first preset formula to calculate the norm corresponding to each of the first complex-valued vectors, wherein the first preset The formula is: ⁇ (w i) is the norm, x is the value of the first complex-valued vectors;
- the second calculation subunit is configured to substitute each of the norms into a second preset formula to calculate the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- the realization process of the functions and roles of the first calculation subunit and the second calculation subunit in the above-mentioned smart interview device is detailed in the realization process corresponding to steps S2221 to S2222 in the above-mentioned text matching-based smart interview method. I won't repeat them here.
- the second calculation unit includes:
- the third calculation subunit is configured to respectively multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain respective weighted word density matrices;
- the combination subunit is used to add the weighted word density matrices to obtain the mixed density matrix.
- the implementation process of the functions and roles of the third calculation subunit and the combination subunit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S2231 to S2232 in the above-mentioned text matching-based intelligent interview method, here No longer.
- calculation sub-module includes:
- x i > tr( ⁇
- the combination unit is used to combine each of the second probabilities to obtain the probability vector.
- the implementation process of the functions and roles of the third computing unit and the combination unit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S231 to S232 in the above-mentioned intelligent interview method based on text matching. Go into details.
- calculation module 3 includes:
- the second calculation sub-module is used to calculate the cosine value between the first vector and the second vector
- the marking module is used to use the cosine value as the similarity.
- the implementation process of the functions and roles of the second calculation sub-module and the marking module in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S31 to S32 in the above-mentioned text-based intelligent interview method. Go into details again.
- This embodiment provides an intelligent interview device based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into a corresponding complex-valued vector, and then the complex-valued vector is calculated correspondingly to obtain the interview text And the standard text respectively corresponding to the mixed density matrix, and select the largest probability vector from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then calculate the difference between the first vector and the second vector Cosine obtains the similarity between the interview text and the standard text, and finally obtains the corresponding interview score according to the similarity matching.
- the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large.
- the range is increased to realize the high accuracy and objectivity of the smart interview.
- an embodiment of the present application also provides a computer device.
- the computer device may be a server, and its internal structure may be as shown in FIG. 3.
- the computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor designed by the computer is used to provide calculation and control capabilities.
- the memory of the computer device includes a non-volatile storage medium and an internal memory.
- the non-volatile storage medium stores an operating system, a computer program, and a database.
- the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
- the database of the computer equipment is used to store data such as standard text.
- the network interface of the computer device is used to communicate with an external terminal through a network connection.
- the computer program is executed by the processor to realize the function of the intelligent interview method based on text matching in any of the above embodiments.
- the steps for the processor to execute the intelligent interview method based on text matching are:
- S1 Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
- S2 Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
- An embodiment of the present application also provides a computer-readable storage medium.
- the storage medium may be a non-volatile storage medium or a volatile storage medium, on which a computer program is stored, and when the computer program is executed by a processor.
- the intelligent interview method based on text matching in any of the above embodiments is specifically as follows:
- S1 Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
- S2 Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
- Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
- Volatile memory may include random access memory (RAM) or external cache memory.
- RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Economics (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The present application provides a text matching-based intelligent interview method and apparatus, a computer device, and a computer readable storage medium, relating to the field of semantic analysis. Said method comprises: acquiring an interview text and a standard text; performing vector transformation on the interview text and the standard text according to a first preset rule, so as to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text; calculating a similarity between the first vector and the second vector; and matching the similarity with the corresponding interview score. In the present application, an interview text and a standard text are processed by means of the described processing to obtain a first vector and a second vector, so that word meanings to be expressed by a text itself can be expressed to the greatest extent, thereby greatly improving the precision of text similarity matching based on a vector level, and achieving the high precision and objectivity of intelligent interview.
Description
本申请要求于2019年10月29日提交中国专利局、申请号为201911037921.4,发明名称为“基于文本匹配的智能面试方法、装置和计算机设备”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on October 29, 2019, the application number is 201911037921.4, and the invention title is "Intelligent Interview Method, Apparatus and Computer Equipment Based on Text Matching", the entire content of which is by reference Incorporated in this application.
本申请涉及人工智能技术领域,特别涉及一种基于文本匹配的智能面试方法、装置和计算机设备。This application relates to the field of artificial intelligence technology, in particular to an intelligent interview method, device and computer equipment based on text matching.
目前各行业中招聘岗位一般是通过人工进行面试,特别是对于某些流动性大、招聘需求大的岗位,人力部门需要花费大量的精力和资源开展频繁的面试。发明人意识到,这种方式会导致招聘的成本较高,需要耗费大量人力资源。并且,人工面试会受到面试官个人的主观意识影响,无法实现对应聘者筛选标准的统一性和客观性。At present, job interviews in various industries are generally conducted manually, especially for certain positions with high mobility and high recruitment demand, the human department needs to spend a lot of energy and resources to conduct frequent interviews. The inventor realizes that this method will lead to higher recruitment costs and require a lot of human resources. In addition, manual interviews will be affected by the personal subjective consciousness of the interviewer, and the unity and objectivity of the screening criteria for candidates cannot be achieved.
本申请的主要目的为提供一种自然语言时间词的解析方法、装置和计算机设备,旨在解决现有时间词解析方法过于呆板和准确率、完整度低的弊端。The main purpose of this application is to provide a natural language time word parsing method, device and computer equipment, aiming to solve the disadvantages of the existing time word parsing method that is too rigid, accurate and low in completeness.
本申请的主要目的为提供一种基于文本匹配的智能面试方法、装置和计算机设备,旨在解决现有面试方法缺乏统一性和客观性的弊端。The main purpose of this application is to provide an intelligent interview method, device and computer equipment based on text matching, aiming to solve the shortcomings of existing interview methods that lack uniformity and objectivity.
为实现上述目的,第一方面,本申请提供了一种基于文本匹配的智能面试方法,包括:In order to achieve the above objectives, in the first aspect, this application provides an intelligent interview method based on text matching, including:
获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;
根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;
根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
第二方面,本申请还提供了一种基于文本匹配的智能面试装置,包括:In the second aspect, this application also provides an intelligent interview device based on text matching, including:
获取模块,用于获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;An acquisition module for acquiring interview text and standard text, where the interview text is the text formed by the interviewer after answering the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
转化模块,用于根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;A conversion module, configured to perform vector conversion on the interview text and the standard text respectively according to a first preset rule to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
计算模块,用于计算所述第一向量和所述第二向量之间的相似度;A calculation module for calculating the similarity between the first vector and the second vector;
匹配模块,用于根据所述相似度匹配对应的面试分数。The matching module is used to match the corresponding interview scores according to the similarity.
第三方面,本申请还提供一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机程序,所述处理器执行所述计算机程序时实现上述基于文本匹配的智能面试方法,其中,所述基于文本匹配的智能面试方法包括以下步骤:In a third aspect, the present application also provides a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned text matching-based intelligent interview method when the computer program is executed, wherein: The intelligent interview method based on text matching includes the following steps:
获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;
根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;
根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
第四方面,本申请还提供一种计算机可读存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时实现上述任一项所述的方法的步骤基于文本匹配的智能面试方法,其中,所述基于文本匹配的智能面试方法包括以下步骤:In a fourth aspect, the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the above methods are realized. A text matching-based intelligent interview method , Wherein the intelligent interview method based on text matching includes the following steps:
获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;
根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;
根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
本申请中提供的一种基于文本匹配的智能面试方法、装置和计算机设备,通过将应聘者作出的面试文本和预先录入的标准文本转化为对应的复值向量后,再将复值向量经过相应的计算得到面试文本和标准文本分别对应的混合密度矩阵,并从各滑动窗口的混合密度矩阵中选出最大概率向量组成各自对应的第一向量和第二向量,然后通过计算第一向量和第二向量之间的余弦得到面试文本和标准文本之间的相似度,最后根据相似度匹配得到相应的面试评分。本申请将面试文本和标准文本通过上述处理后得到的第一向量和第二向量,可以最大程度的表现出文本本身所要表达的词义,从而使得基于向量层面上的文本相似度匹配的精确度大幅度提高,实现智能面试的高精准性和客观性。This application provides an intelligent interview method, device and computer equipment based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into the corresponding complex-valued vector, and then the complex-valued vector is passed through the corresponding The calculation of the interview text and the standard text respectively corresponds to the mixed density matrix, and the maximum probability vector is selected from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then the first vector and the second vector are calculated by The cosine between the two vectors obtains the similarity between the interview text and the standard text, and finally the corresponding interview score is obtained according to the similarity matching. In this application, the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large. The range is improved to realize the high accuracy and objectivity of the smart interview.
图1是本申请一实施例中基于文本匹配的智能面试方法的步骤示意图;FIG. 1 is a schematic diagram of the steps of an intelligent interview method based on text matching in an embodiment of the present application;
图2是本申请一实施例中基于文本匹配的智能面试装置的整体结构框图;Figure 2 is a block diagram of the overall structure of an intelligent interview device based on text matching in an embodiment of the present application;
图3是本申请一实施例的计算机设备的结构示意框图。FIG. 3 is a schematic block diagram of the structure of a computer device according to an embodiment of the present application.
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。The realization, functional characteristics, and advantages of the purpose of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings.
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。In order to make the purpose, technical solutions, and advantages of this application clearer, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application, and are not used to limit the present application.
参照图1,本申请一实施例中提供了一种基于文本匹配的智能面试方法,包括:1, an embodiment of the present application provides an intelligent interview method based on text matching, including:
S1:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;S1: Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
S2:根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;S2: Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
S3:计算所述第一向量和所述第二向量之间的相似度;S3: Calculate the similarity between the first vector and the second vector;
S4:根据所述相似度匹配对应的面试分数。S4: Match the corresponding interview scores according to the similarity.
本实施例中,面试***会根据预先的设置依次输出面试题目给面试者进行回答,面试者可以通过语音或手动输入等方式回答面试题目。面试***在收集到面试者针对面试题目的回答后,形成面试文本,同时调取预先录入的该面试题目对应的标准答案,即标准文本。面试***需要将面试文本和标准文本转化为对应的向量,以便后续进行两者之间的相似度比较,其中,面试***将面试文本和标准文本转化为对应的向量的方法相同,两者的转化动作并没有顺序限制,比如两者可以同时转化,也可以先转化标准文本,再转化面试文本,甚至标准文本可以在进行面试者进行面试前就已经预先转化好后录入面试***中存储,从而使得在面试过程中面试***只需要处理面试文本,加快处理效率。本实施例中以面试文本的向量转化为例进行具体说明,首先,面试***运用独热编码并将面试文本中的各个单词进行复值嵌入(Complex-valued Embedding),使得各个单词产生对应的复值向量,各个复值向量组合形成词矩阵。面试***对词矩阵采用滑动窗口,每次从词矩阵中按照复值向量各自对应的单词在面试文本中的排列顺序,依次递进选择预设数量个复值向量作为第一复值向量并组成第一词矩阵,直至完成对所有复值向量的选择。比如词矩阵中的复值向量为(a,b,c,d,e),预设数量为3,则选择得到的第一词矩阵分别为:(a,b,c),(b,c,d),(c,d,e)。面试***每次对筛选得到的第一词矩阵都需要进行相同的数据处理,具体的处理步骤为:面试***分别将各第一复值向量与各自对应的共轭转置向量相乘得到外积,从而使得由各第一复值向量的外积组成的第一词矩阵转化为词密度矩阵。并且,面试***分别将第 一词矩阵中的各个第一复值向量代入第一预设公式中,计算得到各第一复值向量分别对应的范数。然后,面试***分别将各第一复值向量对应的范数代入第二预设公式中,计算得到各范数分别对应的第一概率。面试***在完成上述的数据处理后,分别将词密度矩阵中的向量乘以各自对应的第一概率,得到计算后的向量组成的各自对应的加权词密度矩阵。面试***将各加权词密度矩阵相加,得到混合密度矩阵。面试***调取第三预设公式,将混合密度矩阵在各个不同的投影平面的投影长度转化为对应的第二概率,从而得到混合密度矩阵投影在各投影平面上的多个第二概率组成的概率向量。其中,第三预设公式为:p
x(p)=<x
i|ρ|x
i>=tr(ρ|x
i><x
i|),其中|x
i>为狄拉克符号表示的词向量,|x
i><x
i|则为|x
i>的外积,px(p)为第二概率,多个概率组成概率向量,i表示第i个投影平面。面试***通过池化操作,分别从各滑动窗口中的多个概率向量中的最大概率向量取出组成第一向量,比如,滑动窗口A的概率向量为(1,2,3),滑动窗口B的概率向量为(4,5,6),滑动窗口C的概率向量为(7,8,9),则分别从滑动窗口A、B、C中选择3、6、9组成第一向量(3,6,9)。面试***按照相同的转化方法,将标准文本转化为对应的第二向量。面试***根据第一向量和第二向量之间的角度计算两者之间的余弦值,计算后的余弦值即为面试文本与标准文本之间的相似度。面试***中预先录入有相似度与面试分数映射关系表,因此面试***可以根据当前次计算得到的相似度得到该面试者在这次面试题目中得到的评价分数。
In this embodiment, the interview system will sequentially output interview questions to the interviewer for answers according to preset settings, and the interviewer can answer the interview questions by voice or manual input. After the interview system collects the interviewer's answers to the interview question, it forms an interview text, and at the same time calls the pre-entered standard answer corresponding to the interview question, that is, the standard text. The interview system needs to convert the interview text and standard text into corresponding vectors in order to compare the similarity between the two later. Among them, the interview system converts the interview text and standard text into corresponding vectors in the same way, and the conversion between the two There is no restriction on the sequence of actions. For example, the two can be converted at the same time, or the standard text can be converted first, and then the interview text can be converted. Even the standard text can be pre-transformed before the interviewer conducts the interview and then entered and stored in the interview system. During the interview process, the interview system only needs to process the interview text to speed up the processing efficiency. In this embodiment, the vector conversion of the interview text is taken as an example for specific description. First, the interview system uses one-hot encoding and performs Complex-valued Embedding on each word in the interview text, so that each word generates a corresponding complex-valued embedding. Value vector, each complex value vector is combined to form a word matrix. The interview system uses a sliding window for the word matrix, and each time from the word matrix, according to the arrangement order of the respective words corresponding to the complex-valued vector in the interview text, a preset number of complex-valued vectors are successively selected as the first complex-valued vector and formed The first word matrix until the selection of all complex-valued vectors is completed. For example, the complex-valued vector in the word matrix is (a, b, c, d, e), and the preset number is 3, then the first word matrix selected are: (a, b, c), (b, c) ,d),(c,d,e). The interview system needs to perform the same data processing on the filtered first word matrix each time. The specific processing steps are: the interview system respectively multiplies each first complex-valued vector with each corresponding conjugate transpose vector to obtain the outer product , So that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix. In addition, the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector. Then, the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm. After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors. The interview system adds up the weighted word density matrix to obtain the mixed density matrix. The interview system calls the third preset formula, and converts the projection length of the mixed density matrix on each different projection plane into the corresponding second probability, thereby obtaining the multiple second probabilities of the mixed density matrix projected on each projection plane. Probability vector. Among them, the third preset formula is: p x (p)=<x i |ρ|x i >=tr(ρ|x i ><x i |), where |x i > is the word represented by the Dirac symbol Vector, |x i ><x i | is the outer product of |x i >, px(p) is the second probability, multiple probabilities form a probability vector, and i represents the i-th projection plane. Through the pooling operation, the interview system takes out the largest probability vector among the multiple probability vectors in each sliding window to form the first vector. For example, the probability vector of sliding window A is (1,2,3), and the probability vector of sliding window B is (1,2,3). The probability vector is (4,5,6), and the probability vector of the sliding window C is (7,8,9). Then select 3, 6, 9 from the sliding windows A, B, and C to form the first vector (3, 6, 9). According to the same conversion method, the interview system converts the standard text into the corresponding second vector. The interview system calculates the cosine value between the first vector and the second vector based on the angle between the two, and the calculated cosine value is the similarity between the interview text and the standard text. The interview system has pre-entered a similarity and interview score mapping relationship table, so the interview system can obtain the interviewer’s evaluation score in this interview question according to the similarity calculated in the current time.
进一步的,所述根据第一预设规则对所述面试文本进行向量转化,得到所述面试文本对应的第一向量的步骤,包括:Further, the step of performing vector transformation on the interview text according to the first preset rule to obtain the first vector corresponding to the interview text includes:
S21:将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;S21: Perform complex-valued embedding of each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;
S22:采用滑动窗口将所述词矩阵转化为混合密度矩阵;S22: Use a sliding window to convert the word matrix into a mixed density matrix;
S23:按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;S23: Calculate according to a preset algorithm to obtain a number of probability vectors corresponding to the mixing density matrix in different sliding windows;
S24:分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。S24: Select the largest probability vector in each of the sliding windows respectively, and each of the largest probability vectors composes the first vector.
本实施例中,面试***运用独热码编码并将面试文本中的各个单词进行复值嵌入,使得各个单词产生对应的复值向量,各个复值向量组合形成词矩阵。其中,本实施例中的复值向量表达式为z=r(cosθ+i sinθ),其好处在于复值向量可以使得词向量相较于传统的实值向量更能表达隐含的词义。使用复值向量表示的词向量不仅只考虑普通的的振幅加减,还会同时考虑它们的相位信息带来的更高阶的语义,既能达到两个词语相加有更多词义的效果,同样也可以使得两个词语相加产生反作用。面试***对词矩阵采用滑动窗口,每次从词矩阵中按照复值向量各自对应的单词在面试文本中的排列顺序,依次递进选择预设数量个复值向量作为第一复值向量并组成第一词矩阵,直至完成对所有复值向量的选择,比如词矩阵中的复值向量为(a,b,c,d,e),预设数量为3,则选择得到的第一词矩阵分别为:(a,b,c),(b,c,d),(c,d,e)。面试***每次对筛选得到的第一词矩阵都需要进行相同的数据处理,具体的处理步骤为:面试***分别将各第一复值向量a与各自对应的共轭转置向量aT相乘得到外积,从而使得由各第一复值向量的外积组成的第一词矩阵转化为词密度矩阵。并且,面试***分别将第一词矩阵中的各个第一复值向量代入第一预设公式中,计算得到各第一复值向量分别对应的范数。然后,面试***分别将各第一复值向量对应的范数代入第二预设公式中,计算得到各范数分别对应的所述第一概率。面试***在完成上述的数据处理后,分别将词密度矩阵中的向量乘以各自对应的第一概率,得到计算后的向量组成的各自对应的加权词密度矩阵。面试***将各加权词密度矩阵相加,得到混合密度矩阵。面试***调取第三预设公式,将混合密度矩阵在各个不同投影平面的投影长度转化为对应的第二概率,从而得到混合密度矩阵投影在各投影平面上的多个第二概率组成的概率向量。面试***通过池化操作,分别从每个滑动窗口的多个概率向量中的最大概率向量取出组成 第一向量。In this embodiment, the interview system uses one-hot code encoding and complex-valued embedding each word in the interview text, so that each word generates a corresponding complex-valued vector, and each complex-valued vector is combined to form a word matrix. Among them, the complex-valued vector expression in this embodiment is z=r(cosθ+i sinθ), and its advantage is that the complex-valued vector can make the word vector more capable of expressing the implicit word meaning than the traditional real-valued vector. The word vectors represented by complex-valued vectors not only consider ordinary amplitude addition and subtraction, but also consider the higher-order semantics brought by their phase information, which can achieve the effect of adding two words to have more meanings. The addition of two words can also be counterproductive. The interview system uses a sliding window for the word matrix, and each time from the word matrix, according to the arrangement order of the respective words corresponding to the complex-valued vector in the interview text, a preset number of complex-valued vectors are successively selected as the first complex-valued vector and formed The first word matrix, until the selection of all complex-valued vectors is completed, for example, the complex-valued vector in the word matrix is (a,b,c,d,e), and the preset number is 3, then select the first word matrix obtained They are: (a, b, c), (b, c, d), (c, d, e). The interview system needs to perform the same data processing on the filtered first word matrix each time. The specific processing steps are: the interview system respectively multiplies each first complex-valued vector a by the corresponding conjugate transpose vector aT to get The outer product, so that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix. In addition, the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector. Then, the interview system respectively substitutes the norm corresponding to each first complex-valued vector into a second preset formula, and calculates the first probability corresponding to each norm. After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors. The interview system adds up the weighted word density matrix to obtain the mixed density matrix. The interview system calls the third preset formula, converts the projection length of the mixed density matrix on each different projection plane into the corresponding second probability, thereby obtaining the probability composed of multiple second probabilities of the mixed density matrix projected on each projection plane vector. Through the pooling operation, the interview system takes out the largest probability vector among the multiple probability vectors in each sliding window to form the first vector.
进一步的,所述采用滑动窗口将所述词矩阵转化为混合密度矩阵的步骤,包括:Further, the step of using a sliding window to convert the word matrix into a mixed density matrix includes:
S221:按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;S221: According to the arrangement order of the words corresponding to each of the complex-valued vectors in the interview text, use the sliding window to sequentially select a preset number of first complex-valued vectors from the word matrix. Matrix, until the selection of all the complex-valued vectors is completed, and a number of first word matrices are obtained;
S222:分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;S222: Calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, convert each of the first word matrices into corresponding word density matrices, and Calculating the first probability corresponding to each of the first complex-valued vectors;
S223:根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。S223: Calculate and obtain a mixed density matrix according to each of the word density matrixes and each of the first probabilities.
本实施例中,面试***采用滑动窗口,每次从词矩阵中按照复值向量各自对应的单词在面试文本中的排列顺序,依次递进选择预设数量个复值向量作为第一复值向量并组成第一词矩阵,直至完成对所有复值向量的选择,比如词矩阵中的复值向量为(a,b,c,d,e),预设数量为3,则选择得到的第一词矩阵分别为:(a,b,c),(b,c,d),(c,d,e)。面试***每次对筛选得到的第一词矩阵都需要进行相同的数据处理,具体的处理步骤为:面试***分别将各第一复值向量a与各自对应的共轭转置向量a
T相乘得到外积,从而使得由各第一复值向量的外积组成的第一词矩阵转化为词密度矩阵。并且,面试***分别将第一词矩阵中的各个第一复值向量代入第一预设公式中,计算得到各第一复值向量分别对应的范数。然后,面试***分别将各第一复值向量对应的范数代入第二预设公式中,计算得到各范数分别对应的第一概率。面试***在完成上述的数据处理后,分别将词密度矩阵中的向量乘以各自对应的第一概率,得到计算后的向量组成的各自对应的加权词密度矩阵。面试***将各加权词密度矩阵相加,得到混合密度矩阵。在应用中,相比于普通的平均加权,本实施例根据局部单词进行加权,使得该***对不同的单词有不同的权重,并且能结合上下文对单词做出判断。
In this embodiment, the interview system uses a sliding window, and each time a preset number of complex-valued vectors are sequentially selected as the first complex-valued vector from the word matrix according to the sequence of the respective words corresponding to the complex-valued vectors in the interview text. And compose the first word matrix until the selection of all the complex-valued vectors is completed. For example, the complex-valued vector in the word matrix is (a,b,c,d,e), and the preset number is 3, then select the first The word matrix is: (a, b, c), (b, c, d), (c, d, e). The interview system needs to perform the same data processing on the filtered first word matrix each time. The specific processing steps are: the interview system respectively multiplies each first complex-valued vector a by the corresponding conjugate transposed vector a T The outer product is obtained, so that the first word matrix composed of the outer product of each first complex-valued vector is transformed into a word density matrix. In addition, the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector. Then, the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm. After completing the above-mentioned data processing, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain their respective weighted word density matrices composed of the calculated vectors. The interview system adds up the weighted word density matrix to obtain the mixed density matrix. In application, compared with ordinary average weighting, this embodiment performs weighting based on local words, so that the system has different weights for different words, and can make judgments on words in combination with context.
进一步的,所述计算各所述第一复值向量分别对应的第一概率的步骤,包括:Further, the step of calculating the first probability corresponding to each of the first complex-valued vectors includes:
S2221:分别将各所述第一复值向量代入第一预设公式中,计算得到各所述第一复值向量分别对应的范数,其中,所述第一预设公式为:
π(w
i)为所述范数,x为所述第一复值向量的值;
S2221: Substitute each of the first complex-valued vectors into a first preset formula, and calculate the norm corresponding to each of the first complex-valued vectors, where the first preset formula is: π (w i) is the norm, x is the value of the first complex-valued vectors;
S2222:分别将各所述范数代入第二预设公式中,计算得到各所述范数分别对应的所述第一概率,其中,所述第二预设公式为:
p(w
i)为所述第一概率,e为自然底数,l表示共有l个w,j表示第j个w。
S2222: Substitute each of the norms into a second preset formula, and calculate the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
本实施例中,面试***分别将第一词矩阵中的各个第一复值向量代入第一预设公式中,计算得到各第一复值向量分别对应的范数,其中,第二预设公式为:
π(w
i)为需要求得的范数,x为第一复值向量的值。然后,面试***分别将各第一复值向量对应的范数代入第二预设公式中,计算得到各范数分别对应的第一概率,其中,第二预设公式为:
p(w
i)为第一概率,e是自然底数,l表示共有l个w,j表示第j个w。
In this embodiment, the interview system respectively substitutes each first complex-valued vector in the first word matrix into the first preset formula, and calculates the norm corresponding to each first complex-valued vector. The second preset formula for: π (w i) is the norm requires determined, x is the value of the first complex-valued vectors. Then, the interview system respectively substitutes the norm corresponding to each first complex-valued vector into the second preset formula, and calculates the first probability corresponding to each norm, where the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
进一步的,所述根据所述词密度矩阵和各所述第一概率计算得到混合密度矩阵的步骤,包括:Further, the step of calculating a mixed density matrix according to the word density matrix and each of the first probabilities includes:
S2231:分别将各所述词密度矩阵中的向量乘以各自对应的所述第一概率,得到各自对应的加权词密度矩阵;S2231: Multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain respective weighted word density matrices;
S2232:将各所述加权词密度矩阵相加,得到所述混合密度矩阵。S2232: Add the weighted word density matrices to obtain the mixed density matrix.
本实施例中,面试***分别将词密度矩阵中的向量乘以各自对应的第一概率,得到计算后的向量组成的各自对应的加权词密度矩阵。面试***将各加权词密度矩阵相加,得到混合密度矩。在应用中,相比于普通的平均加权,本实施例根据局部单词进行加权,使得该***对不同的单词有不同的权重,并且能结合上下文对单词做出判断。In this embodiment, the interview system respectively multiplies the vectors in the word density matrix by their corresponding first probabilities to obtain the respective weighted word density matrices composed of the calculated vectors. The interview system adds up the weighted word density matrix to obtain the mixed density moment. In application, compared with ordinary average weighting, this embodiment performs weighting based on local words, so that the system has different weights for different words, and can make judgments on words in combination with context.
进一步的,所述按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量的步骤,包括:Further, the step of calculating according to a preset algorithm to obtain a plurality of probability vectors corresponding to the mixing density matrix in different sliding windows respectively includes:
S231:将所述混合密度矩阵代入第三预设公式中,计算得到若干个第二概率,其中,所述第三预设公式为:p
x(p)=<x
i|ρ|x
i>=tr(ρ|x
i><x
i|),其中|x
i>初始值为狄拉克符号表示的正交的独热编码向量,|x
i><x
i|则为|x
i>的外积,p
x(p)为第二概率,i表示第i个投影平面;
S231: Substituting the mixed density matrix into a third preset formula to calculate a number of second probabilities, where the third preset formula is: p x (p)=<x i |ρ|x i > = tr (ρ | x i> <x i |), where | x i> orthogonal hot encoded value of the initial vector Dirac notation, | x i><x i | , compared with | x i> a Outer product, p x (p) is the second probability, i represents the i-th projection plane;
S232:将各所述第二概率组合得到所述概率向量。S232: Combine each of the second probabilities to obtain the probability vector.
本实施例中,面试***调取第三预设公式,将混合密度矩阵在投影平面的投影长度转化为对应的第二概率,面试***将得到的各个第二概率组成,从而得到混合密度矩阵在当前滑动窗口对应的投影平面的概率向量。其中,第三预设公式为:p
x(p)=<x
i|ρ|x
i>=tr(ρ|x
i><x
i|),其中|x>初始值为狄拉克符号表示的正交的独热编码向量,在模型训练中长度始终为单位长度,|x
i><x
i|则为|x
i>的外积,用于被训练作为抽取混合密度矩阵的高纬度特征的投影平面,p
x(p)为第二概率。
In this embodiment, the interview system calls the third preset formula, converts the projection length of the mixed density matrix on the projection plane into the corresponding second probability, and the interview system composes the obtained second probabilities to obtain the mixed density matrix in The probability vector of the projection plane corresponding to the current sliding window. Among them, the third preset formula is: p x (p)=<x i |ρ|x i >=tr(ρ|x i ><x i |), where the initial value of |x> is represented by Dirac notation Orthogonal one-hot encoding vector, the length is always unit length in model training, |x i ><x i | is the outer product of |x i >, used to be trained as the high-latitude feature of extracting the mixed density matrix Projection plane, p x (p) is the second probability.
进一步的,所述计算所述第一向量和所述第二向量之间的相似度的步骤,包括:Further, the step of calculating the similarity between the first vector and the second vector includes:
S31:计算所述第一向量和所述第二向量之间的余弦值;S31: Calculate the cosine value between the first vector and the second vector;
S32:将所述余弦值作为所述相似度。S32: Use the cosine value as the similarity.
本实施例中,面试文本转化的第一向量和标准文本转化的第二向量是两个文本各自的N-gram在同一个投影平面投影后的到的向量,因此这两个向量才能具有进行计算的基础。面试***计算第一向量和第二向量之间角度的余弦,计算得到的余弦值即可作为两个文本之间的相似度。比如,a,b为两个向量,则两个向量之间的余弦为:cos(θ)=a×b/(|a|×|b|)。In this embodiment, the first vector of the interview text conversion and the second vector of the standard text conversion are the vectors obtained after the N-grams of the two texts are projected on the same projection plane, so these two vectors can be calculated Foundation. The interview system calculates the cosine of the angle between the first vector and the second vector, and the calculated cosine value can be used as the similarity between the two texts. For example, if a and b are two vectors, the cosine between the two vectors is: cos(θ)=a×b/(|a|×|b|).
本实施例提供的一种基于文本匹配的智能面试方法,通过将应聘者作出的面试文本和预先录入的标准文本转化为对应的复值向量后,再将复值向量经过相应的计算得到面试文本和标准文本分别对应的混合密度矩阵,并从各滑动窗口的混合密度矩阵中选出最大概率向量组成各自对应的第一向量和第二向量,然后通过计算第一向量和第二向量之间的余弦得到面试文本和标准文本之间的相似度,最后根据相似度匹配得到相应的面试评分。本申请将面试文本和标准文本通过上述处理后得到的第一向量和第二向量,可以最大程度的表现出文本本身所要表达的词义,从而使得基于向量层面上的文本相似度匹配的精确度大幅度提高,实现智能面试的高精准性和客观性。This embodiment provides an intelligent interview method based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into a corresponding complex-valued vector, and then the complex-valued vector is calculated correspondingly to obtain the interview text And the standard text respectively corresponding to the mixed density matrix, and select the largest probability vector from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then calculate the difference between the first vector and the second vector Cosine obtains the similarity between the interview text and the standard text, and finally obtains the corresponding interview score according to the similarity matching. In this application, the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large. The range is improved to realize the high accuracy and objectivity of the smart interview.
参照图2,本申请一实施例中还提供了一种基于文本匹配的智能面试装置,包括:2, an embodiment of the present application also provides an intelligent interview device based on text matching, including:
获取模块1,用于获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;The obtaining module 1 is used to obtain interview text and standard text, where the interview text is the text formed by the interviewer after answering the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
转化模块2,用于根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;The conversion module 2 is configured to perform vector conversion on the interview text and the standard text respectively according to a first preset rule to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
计算模块3,用于计算所述第一向量和所述第二向量之间的相似度;The calculation module 3 is used to calculate the similarity between the first vector and the second vector;
匹配模块4,用于根据所述相似度匹配对应的面试分数。The matching module 4 is used to match the corresponding interview scores according to the similarity.
本实施例中,上述智能面试装置中的获取模块1、转化模块2、计算模块3与匹配模块4的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S1至S4的实现过程,在此不再赘述。In this embodiment, the realization process of the functions and functions of the acquisition module 1, the conversion module 2, the calculation module 3 and the matching module 4 in the above-mentioned smart interview device is detailed in the corresponding steps S1 to S4 in the above-mentioned text matching-based smart interview method. The realization process of, I will not repeat it here.
进一步的,所述转化模块2,包括:Further, the conversion module 2 includes:
嵌入子模块,用于将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;The embedding sub-module is used to perform complex-valued embedding of each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;
转化子模块,用于采用滑动窗口将所述词矩阵转化为混合密度矩阵;A transformation sub-module for converting the word matrix into a mixed density matrix by using a sliding window;
第一计算子模块,用于按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;The first calculation sub-module is configured to calculate according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;
选择子模块,用于分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。The selection sub-module is configured to select the largest probability vector in each of the sliding windows, and each of the largest probability vectors composes the first vector.
本实施例中,上述智能面试装置中的嵌入子模块、转化子模块、第一计算子模块与选择子模块的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S21至S24的实现过程,在此不再赘述。In this embodiment, the implementation process of the functions and roles of the embedded sub-module, the conversion sub-module, the first calculation sub-module, and the selection sub-module in the above-mentioned smart interview device is detailed in the corresponding step S21 in the above-mentioned text-based smart interview method. The implementation process to S24 will not be repeated here.
进一步的,所述转化子模块,包括:Further, the transformation sub-module includes:
选择单元,用于按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;The selection unit is configured to use the sliding window to sequentially select a preset number of first complexes from the word matrix according to the order in which the words corresponding to each of the complex-valued vectors are arranged in the interview text. The value vectors form a matrix until the selection of all the complex value vectors is completed, and a number of first word matrices are obtained;
第一计算单元,用于分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;The first calculation unit is configured to calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, and convert each of the first word matrices into corresponding A word density matrix, and calculating the first probability corresponding to each of the first complex-valued vectors;
第二计算单元,用于根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。The second calculation unit is configured to calculate a mixture density matrix according to each of the word density matrix and each of the first probabilities.
本实施例中,上述智能面试装置中的选择单元、第一计算单元与第二计算单元的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S221至S223的实现过程,在此不再赘述。In this embodiment, the implementation process of the functions and roles of the selection unit, the first calculation unit and the second calculation unit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S221 to S223 in the above-mentioned text matching-based intelligent interview method. , I won’t repeat it here.
进一步的,所述第一计算单元,包括:Further, the first calculation unit includes:
第一计算子单元,用于分别将各所述第一复值向量代入第一预设公式中,计算得到各所述第一复值向量分别对应的范数,其中,所述第一预设公式为:
π(w
i)为所述范数,x为所述第一复值向量的值;
The first calculation subunit is configured to substitute each of the first complex-valued vectors into a first preset formula to calculate the norm corresponding to each of the first complex-valued vectors, wherein the first preset The formula is: π (w i) is the norm, x is the value of the first complex-valued vectors;
第二计算子单元,用于分别将各所述范数代入第二预设公式中,计算得到各所述范数分别对应的所述第一概率,其中,所述第二预设公式为:
p(w
i)为所述第一概率,e为自然底数,l表示共有l个w,j表示第j个w。
The second calculation subunit is configured to substitute each of the norms into a second preset formula to calculate the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
本实施例中,上述智能面试装置中的第一计算子单元与第二计算子单元的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S2221至S2222的实现过程,在此不再赘述。In this embodiment, the realization process of the functions and roles of the first calculation subunit and the second calculation subunit in the above-mentioned smart interview device is detailed in the realization process corresponding to steps S2221 to S2222 in the above-mentioned text matching-based smart interview method. I won't repeat them here.
进一步的,所述第二计算单元,包括:Further, the second calculation unit includes:
第三计算子单元,用于分别将各所述词密度矩阵中的向量乘以各自对应的所述第一概率,得到各自对应的加权词密度矩阵;The third calculation subunit is configured to respectively multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain respective weighted word density matrices;
组合子单元,用于将各所述加权词密度矩阵相加,得到所述混合密度矩阵。The combination subunit is used to add the weighted word density matrices to obtain the mixed density matrix.
本实施例中,上述智能面试装置中的第三计算子单元与组合子单元的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S2231至S2232的实现过程,在此不再赘述。In this embodiment, the implementation process of the functions and roles of the third calculation subunit and the combination subunit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S2231 to S2232 in the above-mentioned text matching-based intelligent interview method, here No longer.
进一步的,所述计算子模块,包括:Further, the calculation sub-module includes:
第三计算单元,用于将所述混合密度矩阵代入第三预设公式中,计算得到若干个第二概率,其中,所述第三预设公式为::p
x(p)=<x
i|ρ|x
i>=tr(ρ|x
i><x
i|),其中|x
i>初始值为狄拉克符号表示的正交的独热编码向量,|x
i><x
i|则为|x
i>的外积,p
x(p)为第二概率,i表示第i个投影平面;
The third calculation unit is configured to substitute the mixed density matrix into a third preset formula to calculate a number of second probabilities, where the third preset formula is: p x (p)=<x i |ρ|x i >=tr(ρ|x i ><x i |), where the initial value of |x i > is the orthogonal one-hot encoding vector represented by the Dirac symbol, and |x i ><x i | Is the outer product of |x i >, p x (p) is the second probability, and i represents the i-th projection plane;
组合单元,用于将各所述第二概率组合得到所述概率向量。The combination unit is used to combine each of the second probabilities to obtain the probability vector.
本实施例中,上述智能面试装置中的第三计算单元与组合单元的功能和作用的实现过 程具体详见上述基于文本匹配的智能面试方法中对应步骤S231至S232的实现过程,在此不再赘述。In this embodiment, the implementation process of the functions and roles of the third computing unit and the combination unit in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S231 to S232 in the above-mentioned intelligent interview method based on text matching. Go into details.
进一步的,所述计算模块3,包括:Further, the calculation module 3 includes:
第二计算子模块,用于计算所述第一向量和所述第二向量之间的余弦值;The second calculation sub-module is used to calculate the cosine value between the first vector and the second vector;
标记模块,用于将所述余弦值作为所述相似度。The marking module is used to use the cosine value as the similarity.
本实施例中,上述智能面试装置中的第二计算子模块与标记模块的功能和作用的实现过程具体详见上述基于文本匹配的智能面试方法中对应步骤S31至S32的实现过程,在此不再赘述。In this embodiment, the implementation process of the functions and roles of the second calculation sub-module and the marking module in the above-mentioned intelligent interview device is detailed in the implementation process corresponding to steps S31 to S32 in the above-mentioned text-based intelligent interview method. Go into details again.
本实施例提供的一种基于文本匹配的智能面试装置,通过将应聘者作出的面试文本和预先录入的标准文本转化为对应的复值向量后,再将复值向量经过相应的计算得到面试文本和标准文本分别对应的混合密度矩阵,并从各滑动窗口的混合密度矩阵中选出最大概率向量组成各自对应的第一向量和第二向量,然后通过计算第一向量和第二向量之间的余弦得到面试文本和标准文本之间的相似度,最后根据相似度匹配得到相应的面试评分。本申请将面试文本和标准文本通过上述处理后得到的第一向量和第二向量,可以最大程度的表现出文本本身所要表达的词义,从而使得基于向量层面上的文本相似度匹配的精确度大幅度提高,实现智能面试的高精准性和客观性。This embodiment provides an intelligent interview device based on text matching, which converts the interview text made by the applicant and the pre-entered standard text into a corresponding complex-valued vector, and then the complex-valued vector is calculated correspondingly to obtain the interview text And the standard text respectively corresponding to the mixed density matrix, and select the largest probability vector from the mixed density matrix of each sliding window to form the corresponding first vector and second vector, and then calculate the difference between the first vector and the second vector Cosine obtains the similarity between the interview text and the standard text, and finally obtains the corresponding interview score according to the similarity matching. In this application, the first vector and the second vector obtained from the interview text and the standard text through the above processing can show the word meaning of the text itself to the greatest extent, so that the accuracy of the text similarity matching based on the vector level is large. The range is increased to realize the high accuracy and objectivity of the smart interview.
参照图3,本申请实施例中还提供一种计算机设备,该计算机设备可以是服务器,其内部结构可以如图3所示。该计算机设备包括通过***总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设计的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作***、计算机程序和数据库。该内存储器为非易失性存储介质中的操作***和计算机程序的运行提供环境。该计算机设备的数据库用于存储标准文本等数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现上述的任一实施例基于文本匹配的智能面试方法的功能。Referring to FIG. 3, an embodiment of the present application also provides a computer device. The computer device may be a server, and its internal structure may be as shown in FIG. 3. The computer equipment includes a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor designed by the computer is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used to store data such as standard text. The network interface of the computer device is used to communicate with an external terminal through a network connection. The computer program is executed by the processor to realize the function of the intelligent interview method based on text matching in any of the above embodiments.
上述处理器执行上述基于文本匹配的智能面试方法的步骤为:The steps for the processor to execute the intelligent interview method based on text matching are:
S1:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;S1: Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
S2:根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;S2: Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
S3:计算所述第一向量和所述第二向量之间的相似度;S3: Calculate the similarity between the first vector and the second vector;
S4:根据所述相似度匹配对应的面试分数。S4: Match the corresponding interview scores according to the similarity.
本申请一实施例还提供一种计算机可读存储介质,所述存储介质可以是非易失性存储介质,也可以是易失性存储介质,其上存储有计算机程序,计算机程序被处理器执行时实现上述的任一实施例基于文本匹配的智能面试方法,具体为:An embodiment of the present application also provides a computer-readable storage medium. The storage medium may be a non-volatile storage medium or a volatile storage medium, on which a computer program is stored, and when the computer program is executed by a processor The intelligent interview method based on text matching in any of the above embodiments is specifically as follows:
S1:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;S1: Obtain the interview text and standard text, where the interview text is the text formed by the interviewer's answer to the interview question, and the standard text is the text of the standard answer corresponding to the interview question;
S2:根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;S2: Perform vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;
S3:计算所述第一向量和所述第二向量之间的相似度;S3: Calculate the similarity between the first vector and the second vector;
S4:根据所述相似度匹配对应的面试分数。S4: Match the corresponding interview scores according to the similarity.
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机程序来指令相关的硬件来完成,所述的计算机程序可存储与一非易失性计算机可读取存储介质中,该计算机程序在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的和实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可以包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储 器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM通过多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双速据率SDRAM(SSRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through a computer program. The computer program can be stored and a non-volatile computer readable storage. In the medium, when the computer program is executed, it may include the procedures of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and/or volatile memory. Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not a limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
需要说明的是,在本文中,术语“包括”、“包含”或者其任何其它变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、装置、物品或者方法不仅包括那些要素,而且还包括没有明确列出的其它要素,或者是还包括为这种过程、装置、物品或者方法所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括该要素的过程、装置、物品或者方法中还存在另外的相同要素。It should be noted that in this article, the terms "include", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements not only includes those elements, It also includes other elements not explicitly listed, or elements inherent to the process, device, article, or method. If there are no more restrictions, the element defined by the sentence "including a..." does not exclude the existence of other identical elements in the process, device, article, or method that includes the element.
以上所述仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其它相关的技术领域,均同理包括在本申请的专利保护范围内。The above are only the preferred embodiments of this application, and do not therefore limit the scope of the patent of this application. Any equivalent structure or equivalent process transformation made using the content of the description and drawings of this application, or directly or indirectly applied to other related The technical field is equally included in the scope of patent protection of this application.
Claims (20)
- 一种基于文本匹配的智能面试方法,其中包括:An intelligent interview method based on text matching, including:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
- 根据权利要求1所述的基于文本匹配的智能面试方法,其中,所述根据第一预设规则对所述面试文本进行向量转化,得到所述面试文本对应的第一向量的步骤,包括:The intelligent interview method based on text matching according to claim 1, wherein the step of performing vector transformation on the interview text according to a first preset rule to obtain the first vector corresponding to the interview text comprises:将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;Complex-valued embedding each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;采用滑动窗口将所述词矩阵转化为混合密度矩阵;Using a sliding window to convert the word matrix into a mixed density matrix;按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;Calculating according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。The largest probability vector in each of the sliding windows is selected respectively, and each of the largest probability vectors constitutes the first vector.
- 根据权利要求2所述的基于文本匹配的智能面试方法,其中,所述采用滑动窗口将所述词矩阵转化为混合密度矩阵的步骤,包括:The intelligent interview method based on text matching according to claim 2, wherein the step of using a sliding window to convert the word matrix into a mixed density matrix comprises:按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;According to the arrangement order of the words corresponding to each of the complex-valued vectors in the interview text, the sliding window is used to sequentially select a preset number of first complex-valued vectors from the word matrix to form a matrix, Until the selection of all the complex-valued vectors is completed, a number of first word matrices are obtained;分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;Calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, convert each of the first word matrices into corresponding word density matrices, and calculate each The first probabilities respectively corresponding to the first complex-valued vectors;根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。A mixture density matrix is calculated according to each of the word density matrix and each of the first probabilities.
- 根据权利要求3所述的基于文本匹配的智能面试方法,其中,所述计算各所述第一复值向量分别对应的第一概率的步骤,包括:The intelligent interview method based on text matching according to claim 3, wherein the step of calculating the first probability corresponding to each of the first complex-valued vectors respectively comprises:分别将各所述第一复值向量代入第一预设公式中,计算得到各所述第一复值向量分别对应的范数,其中,所述第一预设公式为: π(w i)为所述范数,x为所述第一复值向量的值; Substituting each of the first complex-valued vectors into a first preset formula respectively, and calculating the norm corresponding to each of the first complex-valued vectors, wherein the first preset formula is: π (w i) is the norm, x is the value of the first complex-valued vectors;分别将各所述范数代入第二预设公式中,计算得到各所述范数分别对应的所述第一概率,其中,所述第二预设公式为: p(w i)为所述第一概率,e为自然底数,l表示共有l个w,j表示第j个w。 Substituting each of the norms into a second preset formula, and calculating the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- 根据权利要求3所述的基于文本匹配的智能面试方法,其中,所述根据所述词密度矩阵和各所述第一概率计算得到混合密度矩阵的步骤,包括:The intelligent interview method based on text matching according to claim 3, wherein the step of calculating a mixture density matrix according to the word density matrix and each of the first probabilities comprises:分别将各所述词密度矩阵中的向量乘以各自对应的所述第一概率,得到各自对应的加权词密度矩阵;Respectively multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain the respective weighted word density matrices;将各所述加权词密度矩阵相加,得到所述混合密度矩阵。The weighted word density matrix is added to obtain the mixed density matrix.
- 根据权利要求2所述的基于文本匹配的智能面试方法,其中,所述按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量的步骤,包括:The intelligent interview method based on text matching according to claim 2, wherein the step of calculating according to a preset algorithm a number of probability vectors corresponding to the mixing density matrix in different sliding windows respectively comprises:将所述混合密度矩阵代入第三预设公式中,计算得到若干个第二概率,其中,所述第三预设公式为:p x(p)=<x i|ρ|x i>=tr(ρ|x i><x i|),其中|x i>初始值为狄拉克符号表示的正交的 独热编码向量,|x i><x i|则为|x i>的外积,p x(p)为第二概率,i表示第i个投影平面; Substituting the mixture density matrix into a third preset formula, and calculating a number of second probabilities, where the third preset formula is: p x (p)=<x i |ρ|x i >=tr (ρ|x i ><x i |), where the initial value of |x i > is the orthogonal one-hot encoding vector represented by the Dirac symbol, and |x i ><x i | is the outer product of |x i> , P x (p) is the second probability, and i represents the i-th projection plane;将各所述第二概率组合得到所述概率向量。Combining each of the second probabilities to obtain the probability vector.
- 根据权利要求1所述的基于文本匹配的智能面试方法,其中,所述计算所述第一向量和所述第二向量之间的相似度的步骤,包括:The intelligent interview method based on text matching according to claim 1, wherein the step of calculating the similarity between the first vector and the second vector comprises:计算所述第一向量和所述第二向量之间的余弦值;Calculating the cosine value between the first vector and the second vector;将所述余弦值作为所述相似度。The cosine value is used as the similarity.
- 一种基于文本匹配的智能面试装置,其中包括:An intelligent interview device based on text matching, which includes:获取模块,用于获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;An acquisition module for acquiring interview text and standard text, where the interview text is the text formed by the interviewer after answering the interview question, and the standard text is the text of the standard answer corresponding to the interview question;转化模块,用于根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;A conversion module, configured to perform vector conversion on the interview text and the standard text respectively according to a first preset rule to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;计算模块,用于计算所述第一向量和所述第二向量之间的相似度;A calculation module for calculating the similarity between the first vector and the second vector;匹配模块,用于根据所述相似度匹配对应的面试分数。The matching module is used to match the corresponding interview scores according to the similarity.
- 根据权利要求8所述的基于文本匹配的智能面试装置,其中,所述转化模块,包括:The intelligent interview device based on text matching according to claim 8, wherein the conversion module comprises:嵌入子模块,用于将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;The embedding sub-module is used to perform complex-valued embedding of each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;转化子模块,用于采用滑动窗口将所述词矩阵转化为混合密度矩阵;A transformation sub-module for converting the word matrix into a mixed density matrix by using a sliding window;第一计算子模块,用于按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;The first calculation sub-module is configured to calculate according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;选择子模块,用于分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。The selection sub-module is configured to select the largest probability vector in each of the sliding windows, and each of the largest probability vectors composes the first vector.
- 根据权利要求9所述的基于文本匹配的智能面试装置,其中,所述转化子模块,包括:The intelligent interview device based on text matching according to claim 9, wherein the conversion sub-module comprises:选择单元,用于按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;The selection unit is configured to use the sliding window to sequentially select a preset number of first complexes from the word matrix according to the order in which the words corresponding to each of the complex-valued vectors are arranged in the interview text. The value vectors form a matrix until the selection of all the complex value vectors is completed, and a number of first word matrices are obtained;第一计算单元,用于分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;The first calculation unit is configured to calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, and convert each of the first word matrices into corresponding A word density matrix, and calculating the first probability corresponding to each of the first complex-valued vectors;第二计算单元,用于根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。The second calculation unit is configured to calculate a mixture density matrix according to each of the word density matrix and each of the first probabilities.
- 一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机程序,其中,所述处理器执行所述计算机程序时实现一种基于文本匹配的智能面试方法:A computer device includes a memory and a processor, and a computer program is stored in the memory, wherein the processor implements an intelligent interview method based on text matching when the processor executes the computer program:其中,所述基于文本匹配的智能面试方法包括:Wherein, the intelligent interview method based on text matching includes:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
- 根据权利要求11所述的计算机设备,其中,所述根据第一预设规则对所述面试文本进行向量转化,得到所述面试文本对应的第一向量的步骤,包括:11. The computer device according to claim 11, wherein the step of performing vector transformation on the interview text according to a first preset rule to obtain the first vector corresponding to the interview text comprises:将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;Complex-valued embedding each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;采用滑动窗口将所述词矩阵转化为混合密度矩阵;Using a sliding window to convert the word matrix into a mixed density matrix;按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;Calculating according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。The largest probability vector in each of the sliding windows is selected respectively, and each of the largest probability vectors constitutes the first vector.
- 根据权利要求12所述的计算机设备,其中,所述采用滑动窗口将所述词矩阵转化为混合密度矩阵的步骤,包括:The computer device according to claim 12, wherein the step of using a sliding window to convert the word matrix into a mixed density matrix comprises:按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;According to the arrangement order of the words corresponding to each of the complex-valued vectors in the interview text, the sliding window is used to sequentially select a preset number of first complex-valued vectors from the word matrix to form a matrix, Until the selection of all the complex-valued vectors is completed, a number of first word matrices are obtained;分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;Calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, convert each of the first word matrices into corresponding word density matrices, and calculate each The first probabilities respectively corresponding to the first complex-valued vectors;根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。A mixture density matrix is calculated according to each of the word density matrix and each of the first probabilities.
- 根据权利要求13所述的计算机设备,其中,所述计算各所述第一复值向量分别对应的第一概率的步骤,包括:The computer device according to claim 13, wherein the step of calculating the first probability corresponding to each of the first complex-valued vectors respectively comprises:分别将各所述第一复值向量代入第一预设公式中,计算得到各所述第一复值向量分别对应的范数,其中,所述第一预设公式为: π(w i)为所述范数,x为所述第一复值向量的值; Substituting each of the first complex-valued vectors into a first preset formula respectively, and calculating the norm corresponding to each of the first complex-valued vectors, wherein the first preset formula is: π (w i) is the norm, x is the value of the first complex-valued vectors;分别将各所述范数代入第二预设公式中,计算得到各所述范数分别对应的所述第一概率,其中,所述第二预设公式为: p(w i)为所述第一概率,e为自然底数,l表示共有l个w,j表示第j个w。 Substituting each of the norms into a second preset formula, and calculating the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- 根据权利要求13所述的计算机设备,其中,所述根据所述词密度矩阵和各所述第一概率计算得到混合密度矩阵的步骤,包括:11. The computer device according to claim 13, wherein the step of calculating a mixture density matrix according to the word density matrix and each of the first probabilities comprises:分别将各所述词密度矩阵中的向量乘以各自对应的所述第一概率,得到各自对应的加权词密度矩阵;Respectively multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain the respective weighted word density matrices;将各所述加权词密度矩阵相加,得到所述混合密度矩阵。The weighted word density matrix is added to obtain the mixed density matrix.
- 一种计算机可读存储介质,其上存储有计算机程序,其中,所述计算机程序被处理器执行时实现一种基于文本匹配的智能面试方法,其中,所述基于文本匹配的智能面试方法包括以下步骤:A computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to realize an intelligent interview method based on text matching, wherein the intelligent interview method based on text matching includes the following step:获取面试文本和标准文本,其中所述面试文本为面试者针对面试题目的回答后形成的文本,所述标准文本为面试题目对应的标准答案的文本;Obtain an interview text and a standard text, where the interview text is a text formed after the interviewer answers to the interview question, and the standard text is a text of a standard answer corresponding to the interview question;根据第一预设规则分别对所述面试文本和所述标准文本进行向量转化,得到所述面试文本对应的第一向量和所述标准文本对应的第二向量;Performing vector transformation on the interview text and the standard text respectively according to a first preset rule, to obtain a first vector corresponding to the interview text and a second vector corresponding to the standard text;计算所述第一向量和所述第二向量之间的相似度;Calculating the similarity between the first vector and the second vector;根据所述相似度匹配对应的面试分数。Match the corresponding interview scores according to the similarity.
- 根据权利要求16所述的计算机可读存储介质,其中,所述根据第一预设规则对所述面试文本进行向量转化,得到所述面试文本对应的第一向量的步骤,包括:15. The computer-readable storage medium according to claim 16, wherein the step of performing vector transformation on the interview text according to a first preset rule to obtain the first vector corresponding to the interview text comprises:将所述面试文本中的各个单词进行复值嵌入,得到各所述单词分别对应的复值向量组成的词矩阵;Complex-valued embedding each word in the interview text to obtain a word matrix composed of complex-valued vectors corresponding to each of the words;采用滑动窗口将所述词矩阵转化为混合密度矩阵;Using a sliding window to convert the word matrix into a mixed density matrix;按照预设算法计算得到所述混合密度矩阵在不同的所述滑动窗口分别对应的若干个概率向量;Calculating according to a preset algorithm several probability vectors corresponding to the mixing density matrix in different sliding windows;分别选择各个所述滑动窗口中最大的概率向量,各所述最大的概率向量组成所述第一向量。The largest probability vector in each of the sliding windows is selected respectively, and each of the largest probability vectors constitutes the first vector.
- 根据权利要求17所述的计算机可读存储介质,其中,所述采用滑动窗口将所述词矩阵转化为混合密度矩阵的步骤,包括:18. The computer-readable storage medium according to claim 17, wherein the step of using a sliding window to convert the word matrix into a mixed density matrix comprises:按照各所述复值向量分别对应的所述单词在所述面试文本中的排列顺序,采用所述滑动窗口从所述词矩阵中依次递进选择预设数量个第一复值向量组成矩阵,直至完成对所有所述复值向量的选择,得到若干个第一词矩阵;According to the arrangement order of the words corresponding to each of the complex-valued vectors in the interview text, the sliding window is used to sequentially select a preset number of first complex-valued vectors from the word matrix to form a matrix, Until the selection of all the complex-valued vectors is completed, a number of first word matrices are obtained;分别计算各所述第一词矩阵中的各所述第一复值向量与各自对应的共轭装置向量的外积,将各所述第一词矩阵转化为对应的词密度矩阵,并计算各所述第一复值向量分别对应的第一概率;Calculate the outer products of each of the first complex-valued vectors in each of the first word matrices and their corresponding conjugate device vectors, convert each of the first word matrices into corresponding word density matrices, and calculate each The first probabilities respectively corresponding to the first complex-valued vectors;根据各所述词密度矩阵和各所述第一概率计算得到混合密度矩阵。A mixture density matrix is calculated according to each of the word density matrix and each of the first probabilities.
- 根据权利要求18所述的计算机可读存储介质,其中,所述计算各所述第一复值向量分别对应的第一概率的步骤,包括:18. The computer-readable storage medium according to claim 18, wherein the step of calculating the first probability corresponding to each of the first complex-valued vectors comprises:分别将各所述第一复值向量代入第一预设公式中,计算得到各所述第一复值向量分别对应的范数,其中,所述第一预设公式为: π(w i)为所述范数,x为所述第一复值向量的值; Substituting each of the first complex-valued vectors into a first preset formula respectively, and calculating the norm corresponding to each of the first complex-valued vectors, wherein the first preset formula is: π (w i) is the norm, x is the value of the first complex-valued vectors;分别将各所述范数代入第二预设公式中,计算得到各所述范数分别对应的所述第一概率,其中,所述第二预设公式为: p(w i)为所述第一概率,e为自然底数,l表示共有l个w,j表示第j个w。 Substituting each of the norms into a second preset formula, and calculating the first probability corresponding to each of the norms, wherein the second preset formula is: p(w i ) is the first probability, e is the natural base, l represents a total of l w, and j represents the j-th w.
- 根据权利要求18所述的计算机可读存储介质,其中,所述根据所述词密度矩阵和各所述第一概率计算得到混合密度矩阵的步骤,包括:18. The computer-readable storage medium according to claim 18, wherein the step of calculating a mixture density matrix according to the word density matrix and each of the first probabilities comprises:分别将各所述词密度矩阵中的向量乘以各自对应的所述第一概率,得到各自对应的加权词密度矩阵;Respectively multiply the vectors in each of the word density matrices by the corresponding first probabilities to obtain the respective weighted word density matrices;将各所述加权词密度矩阵相加,得到所述混合密度矩阵。The weighted word density matrix is added to obtain the mixed density matrix.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911037921.4A CN111027305B (en) | 2019-10-29 | 2019-10-29 | Text matching-based intelligent interview method and device and computer equipment |
CN201911037921.4 | 2019-10-29 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021082488A1 true WO2021082488A1 (en) | 2021-05-06 |
Family
ID=70204652
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2020/098796 WO2021082488A1 (en) | 2019-10-29 | 2020-06-29 | Text matching-based intelligent interview method and apparatus, and computer device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN111027305B (en) |
WO (1) | WO2021082488A1 (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111027305B (en) * | 2019-10-29 | 2023-12-19 | 平安科技(深圳)有限公司 | Text matching-based intelligent interview method and device and computer equipment |
CN111694936B (en) * | 2020-04-26 | 2023-06-06 | 平安科技(深圳)有限公司 | Method, device, computer equipment and storage medium for identification of AI intelligent interview |
CN111798217B (en) * | 2020-07-10 | 2024-06-28 | 河北冀联人力资源服务集团有限公司 | Data analysis system and method |
CN113378065B (en) * | 2021-07-09 | 2023-07-04 | 小红书科技有限公司 | Method for determining content diversity based on sliding spectrum decomposition and method for selecting content |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100179933A1 (en) * | 2009-01-12 | 2010-07-15 | Nec Laboratories America, Inc. | Supervised semantic indexing and its extensions |
CN107818164A (en) * | 2017-11-02 | 2018-03-20 | 东北师范大学 | A kind of intelligent answer method and its system |
CN109978339A (en) * | 2019-02-27 | 2019-07-05 | 平安科技(深圳)有限公司 | AI interviews model training method, device, computer equipment and storage medium |
CN111027305A (en) * | 2019-10-29 | 2020-04-17 | 平安科技(深圳)有限公司 | Intelligent interviewing method and device based on text matching and computer equipment |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10607188B2 (en) * | 2014-03-24 | 2020-03-31 | Educational Testing Service | Systems and methods for assessing structured interview responses |
US10796217B2 (en) * | 2016-11-30 | 2020-10-06 | Microsoft Technology Licensing, Llc | Systems and methods for performing automated interviews |
CN110347787B (en) * | 2019-06-12 | 2022-12-09 | 平安科技(深圳)有限公司 | Interview method and device based on AI auxiliary interview scene and terminal equipment |
-
2019
- 2019-10-29 CN CN201911037921.4A patent/CN111027305B/en active Active
-
2020
- 2020-06-29 WO PCT/CN2020/098796 patent/WO2021082488A1/en active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100179933A1 (en) * | 2009-01-12 | 2010-07-15 | Nec Laboratories America, Inc. | Supervised semantic indexing and its extensions |
CN107818164A (en) * | 2017-11-02 | 2018-03-20 | 东北师范大学 | A kind of intelligent answer method and its system |
CN109978339A (en) * | 2019-02-27 | 2019-07-05 | 平安科技(深圳)有限公司 | AI interviews model training method, device, computer equipment and storage medium |
CN111027305A (en) * | 2019-10-29 | 2020-04-17 | 平安科技(深圳)有限公司 | Intelligent interviewing method and device based on text matching and computer equipment |
Also Published As
Publication number | Publication date |
---|---|
CN111027305A (en) | 2020-04-17 |
CN111027305B (en) | 2023-12-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021082488A1 (en) | Text matching-based intelligent interview method and apparatus, and computer device | |
Li et al. | Visual question generation as dual task of visual question answering | |
CN110287285B (en) | Method and device for identifying problem intention, computer equipment and storage medium | |
CN110795552B (en) | Training sample generation method and device, electronic equipment and storage medium | |
CN109271646A (en) | Text interpretation method, device, readable storage medium storing program for executing and computer equipment | |
CN109859760A (en) | Phone robot voice recognition result bearing calibration based on deep learning | |
CN111061847A (en) | Dialogue generation and corpus expansion method and device, computer equipment and storage medium | |
CN113157863B (en) | Question-answer data processing method, device, computer equipment and storage medium | |
WO2020151310A1 (en) | Text generation method and device, computer apparatus, and medium | |
CN111078847A (en) | Power consumer intention identification method and device, computer equipment and storage medium | |
CN112348362A (en) | Method, device, equipment and medium for determining position candidate | |
CN112365385B (en) | Knowledge distillation method and device based on self attention and computer equipment | |
CN112000788B (en) | Data processing method, device and computer readable storage medium | |
CN108776832A (en) | Information processing method, device, computer equipment and storage medium | |
WO2021003813A1 (en) | Answer generation method based on neural network model, and related device | |
CN111737486A (en) | Human-sentry matching method and storage device based on knowledge graph and deep learning | |
CN113886550A (en) | Question-answer matching method, device, equipment and storage medium based on attention mechanism | |
CN116775843A (en) | Question-answer pair evaluation data generation method, question-answer pair evaluation data generation device, computer equipment and storage medium | |
CN114780723B (en) | Portrayal generation method, system and medium based on guide network text classification | |
CN113204630A (en) | Text matching method and device, computer equipment and readable storage medium | |
CN109840506B (en) | Method for solving video question-answering task by utilizing video converter combined with relational interaction | |
CN116975288A (en) | Text processing method and text processing model training method | |
CN114169447A (en) | Event detection method based on self-attention convolution bidirectional gating cyclic unit network | |
CN113792120B (en) | Graph network construction method and device, reading and understanding method and device | |
CN116127981A (en) | Semantic vector representation method, semantic vector representation device, computer equipment and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20881834 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20881834 Country of ref document: EP Kind code of ref document: A1 |