CN107609101B - Intelligent interaction method, equipment and storage medium - Google Patents
Intelligent interaction method, equipment and storage medium Download PDFInfo
- Publication number
- CN107609101B CN107609101B CN201710815146.5A CN201710815146A CN107609101B CN 107609101 B CN107609101 B CN 107609101B CN 201710815146 A CN201710815146 A CN 201710815146A CN 107609101 B CN107609101 B CN 107609101B
- Authority
- CN
- China
- Prior art keywords
- user
- similarity
- semantic
- question
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses an intelligent interaction method, equipment and a storage medium. The method comprises the following steps: receiving a user question; calculating the similarity between the user question and a pre-stored question in a knowledge base; if the pre-stored problems with the similarity exceeding the threshold exist, outputting answers corresponding to the pre-stored problems in the knowledge base according to the sequence of the similarity; adjusting the calculation mode of the similarity according to the selection record of the user on the output answer; if the pre-stored problem that the similarity exceeds the threshold value does not exist, performing statement adjustment on the spoken language expression of the user problem, and recalculating the similarity; and if the similarity does not exceed the threshold value, performing semantic analysis on the user question, and searching answers related to semantic results from a knowledge base or the Internet. According to the scheme, the accuracy and timeliness of intelligent reply can be improved, and the reliability of intelligent interaction is further provided.
Description
Technical Field
The present application relates to the field of data processing, and in particular, to an intelligent interaction method, device, and storage medium.
Background
With the continuous development of computers and the internet, people's lives have gradually entered the intelligent era. Namely, intelligent devices such as computers, mobile phones, tablet computers and the like can be intelligently interacted with people, and convenient and fast services are provided for various aspects of life of people.
Generally, the smart device needs to perform semantic parsing on information input by a user, and then perform related operations according to a result of the semantic parsing, for example, provide corresponding answers. However, the meaning of the same question or operation command is different due to different expressions or even different moods of people. At present, the intelligent device still cannot provide corresponding reply or accurate reply due to the fact that the meaning of the natural language input by the user cannot be recognized correctly through voice. Therefore, improving the accuracy and timeliness of intelligent reply is a main subject of intelligent interaction at present.
Disclosure of Invention
The technical problem mainly solved by the application is to provide an intelligent interaction method, equipment and a storage medium, which can improve the accuracy and timeliness of intelligent reply and further provide the reliability of intelligent interaction.
In order to solve the above problem, a first aspect of the present application provides an intelligent interaction method, including: receiving a user question; calculating the similarity between the user question and a pre-stored question in a knowledge base; if the pre-stored questions with the similarity exceeding the threshold value with the user question exist in the knowledge base, outputting answers corresponding to the pre-stored questions with the similarity exceeding the threshold value in the knowledge base according to the sequence of the similarity; adjusting the calculation mode of the similarity according to the selection record of the user on the output answer, so that the similarity of the pre-stored questions corresponding to the user question and the answer selected by the user is the highest; if the knowledge base does not have the pre-stored problem of which the similarity with the user problem exceeds a threshold value, performing statement adjustment on the spoken language expression of the user problem, and recalculating the similarity between the adjusted user problem and the pre-stored problem in the knowledge base; and if the similarity does not exceed the threshold value, performing semantic analysis on the user question to obtain a semantic result, and searching answers related to the semantic result from the knowledge base or the Internet.
In order to solve the above problem, a second aspect of the present application provides an intelligent interactive device, comprising a memory and a processor connected to each other; the processor is configured to perform the method described above.
In order to solve the above problem, a third aspect of the present application provides a non-volatile storage medium storing a computer program for execution by a processor to perform the above method.
In the above scheme, the intelligent interaction device outputs the answer corresponding to the pre-stored question by calculating the similarity between the user question and the pre-stored question, and outputs the highest similarity between the user question and the pre-stored question corresponding to the answer selected by the user according to the calculation mode of the similarity after the user selects and records the output answer, so that the calculation of the subsequent similarity can be ensured to be matched with the expression habit of the user, the accuracy of intelligent response is improved, and the reliability of intelligent interaction is improved; when the pre-stored problem with the similarity exceeding the threshold value is not found, sentence adjustment is carried out on the spoken expression of the user problem, and the similarity between the adjusted user problem and the pre-stored problem in the knowledge base is recalculated; if the similarity does not exceed the threshold, the semantic analysis is carried out on the user question, and the answer related to the semantic result is searched from the knowledge base or the Internet, so that the response is timely and accurate, and the accuracy of the semantic analysis can be improved by adjusting the spoken language expression, so that the accuracy of the intelligent response is improved, and the reliability of the intelligent interaction is also improved.
Drawings
FIG. 1 is a flow chart of an embodiment of the intelligent interaction method of the present application;
FIG. 2 is a partial flow diagram of another embodiment of the intelligent interaction method of the present application;
FIG. 3 is a partial flow chart of yet another embodiment of the intelligent interaction method of the present application
FIG. 4 is a schematic structural diagram of an embodiment of an intelligent interaction device according to the present application;
FIG. 5 is a schematic structural diagram of an embodiment of a non-volatile storage medium according to the present application.
Detailed Description
The following describes in detail the embodiments of the present application with reference to the drawings attached hereto.
In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular system structures, interfaces, techniques, etc. in order to provide a thorough understanding of the present application.
The terms "system" and "network" are often used interchangeably herein. The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship.
Referring to fig. 1, fig. 1 is a flowchart illustrating an intelligent interaction method according to an embodiment of the present application. The method is executed by intelligent interaction equipment with processing capacity, such as a terminal or a server of a computer, a mobile phone and the like. In this embodiment, the method includes the steps of:
s110: a user question is received.
The intelligent interactive device may obtain the information input by the user through the internet, for example, the intelligent interactive device is a server which obtains the information input by the user through the user terminal through the internet. Or the intelligent interactive device directly obtains the information input by the user through the input device of the intelligent interactive device.
And particularly, the intelligent interactive device can receive voice information and text information input by a user. And, the voice information and the text information can be received and processed simultaneously. Or, the intelligent interactive device only receives text information or voice information input by the user. When the intelligent interaction equipment receives the voice information, the voice information is subjected to voice recognition to obtain corresponding text information.
S120: and calculating the similarity between the user question and a pre-stored question in a knowledge base.
Specifically, the intelligent interaction device is provided with a knowledge base, and a plurality of pre-stored questions and corresponding answer knowledge points are stored in the knowledge base. When similarity operation is carried out, the intelligent interaction equipment can adopt a shift algorithm to calculate the Jaccard coefficients of the user problems and the pre-stored problems in the knowledge base.
After traversing the pre-stored problems in the knowledge base to calculate the similarity between each pre-stored problem and the user problem, judging whether the pre-stored problem with the similarity exceeding a threshold exists or not, and correspondingly executing the following steps according to the judgment result. The threshold value can be set by a user or intelligent interaction equipment according to a set algorithm according to actual conditions.
In a specific application, the intelligent interaction device can determine the multi-dimensional similarity ranking of the user questions and the pre-stored questions in the knowledge base based on different keywords of the user questions, and synthesize the similarity ranking in each dimension to obtain the similarity between the user questions and the pre-stored questions in the knowledge base. For example, the text information is segmented, specifically, the text information is segmented according to at least one of the position of the user, the service scene of the user and the language habit of the user, at least one keyword in the user problem is selected from the segmentation result, and the similarity calculation is performed on the pre-stored problems according to different keywords or keyword combinations to obtain different similarity ranks. And weighting and summing the sequence numbers or similarity values of the pre-stored problems in the obtained different similarity sequences, and taking the numerical value obtained after weighting and summing as the similarity between the pre-stored problems and the user problems.
S130: and judging whether a pre-stored problem with the similarity exceeding a threshold value with the user problem exists in the knowledge base, if so, executing S140, otherwise, executing S160.
S140: and outputting answers corresponding to the pre-stored questions in the knowledge base according to the similarity sequence.
In this embodiment, the answer corresponding to the pre-stored question with the similarity of the user question exceeding the threshold is obtained from the knowledge base, and the obtained answers are output according to the similarity obtained by the calculation in S120, where the specific output may be represented by displaying the obtained answer on the intelligent interactive device.
In another embodiment, the intelligent interaction device may further calculate a degree of association between the user question and a pre-stored question in the knowledge base, and output an answer corresponding to the pre-stored question with the degree of similarity exceeding a threshold value according to the corresponding degree of association.
S150: and adjusting the calculation mode of the similarity according to the selection record of the user for the output answer, so that the similarity of the user question and the pre-stored question corresponding to the answer selected by the user is the highest.
In order to ensure that subsequent similarity calculation can be matched according to user habits, the intelligent interaction equipment has self-learning capability, after relevant answers are output, the operation of the user on the relevant answers is detected, if some output answers are clicked for checking, forwarding or other operation showing the attention of the user on the answers, the answers are determined to be selected by the user and recorded, according to the selection record of the user on the output answers, pre-stored questions corresponding to the selected output answers are obtained, the obtained pre-stored questions are determined to be questions matched with the semantics of the user questions, therefore, the expression habits of the user on the questions can be analyzed and obtained, and the subsequent similarity calculation mode is adjusted according to the expression habits, and if the adjusted similarity calculation mode is used, the similarity between the user questions and the obtained pre-stored questions is the highest.
S160: and performing statement adjustment on the spoken expression of the user problem, and recalculating the similarity between the adjusted user problem and a pre-stored problem in a knowledge base.
In this embodiment, the intelligent interaction device first determines whether the user question includes a spoken expression, and specifically may compare the user question with a standard question in a knowledge base to determine whether the user question includes a spoken expression. If a spoken expression is included, the spoken words belonging to the spoken expression in the user question may be subject to spoken correction, which may include any one or any combination of word order reversal, deletion, and substitution. For example, if there are two consecutive words with reversed word order in the spoken expression included in the user question, the two consecutive words with reversed word order may be reordered to form a new word. For another example, if the user question includes a spoken language word, the spoken language word is deleted.
After the sentence adjustment is performed, the adjusted user question expression is calculated. Specifically, the similarity may be calculated in the manner described in S120.
S170: and if the similarity does not exceed the threshold value, performing semantic analysis on the user question to obtain a semantic result, and searching answers related to the semantic result from the knowledge base or the Internet.
For example, if the adjusted user question still cannot find a pre-stored question with a similarity satisfying a threshold, performing semantic analysis on the user question to obtain a semantic result, searching for an answer related to the semantic result from the knowledge base or the internet, and outputting the searched answer, where a plurality of answers are found approximately, the searched answers may be sequentially output according to the correlation with the semantic result, and the answers may be sequentially output on an output device of the intelligent interaction device, such as a display screen.
In a specific application, the intelligent interactive device can be used for instant messaging, and the instant messaging comprises WeChat, QQ, mailbox, forum and the like. The instant messaging divides the field for the user in advance, and records the questions answered by the user. The searching for the answer related to the semantic result from the internet in S170 includes: sending the user question to a user in the field of the question or a user who has answered other questions containing the keywords of the user question through instant messaging, and requiring the user to answer in a limited time; and when receiving the reply of the instant messaging user, returning the user reply obtained through the instant messaging.
In another embodiment, after searching for the answer related to the semantic result, the intelligent terminal stores the user question and the searched related answer in the knowledge base to serve as a new pre-stored question and a new related answer in the knowledge base.
In another embodiment, the intelligent interactive device can also input prompt information to the user according to the detected emotional condition of the user. Wherein, the emotional condition of the user is determined according to the speed of speech or typing speed of the user and the input keywords. For example, the intelligent interactive device stores the speech speed, typing speed and keywords corresponding to different emotions in advance. The current user emotion is determined by detecting the speed (the speed of speech and/or the typing speed) when the user inputs natural language and key words in text information input by the user, and prompt information related to the user emotion is input, for example, the current user emotion is angry, and then some comfort prompt information is selected to display the user or to play pleasure music. Further, the intelligent interactive device may also use the emotional condition of the user as the scene information described in the next embodiment to determine the current semantic scene. Moreover, the intelligent interaction device may further select an operation corresponding to the semantic result in combination with the user emotion condition, for example, if the operation determined according to the semantic result is to query a weather forecast, and if the current user emotion is angry, the preset tone corresponding to the emotion is selected to play the weather forecast.
In the embodiment, the intelligent interaction device outputs the answer corresponding to the pre-stored question by calculating the similarity between the user question and the pre-stored question, and outputs the highest similarity between the user question and the pre-stored question corresponding to the answer selected by the user according to the calculation mode of the similarity after the selection record of the user on the output answer is adjusted, so that the calculation of the subsequent similarity can be ensured to be matched with the expression habit of the user, the accuracy of intelligent response is improved, and the reliability of intelligent interaction is improved; when the pre-stored problem with the similarity exceeding the threshold value is not found, sentence adjustment is carried out on the spoken expression of the user problem, and the similarity between the adjusted user problem and the pre-stored problem in the knowledge base is recalculated; if the similarity does not exceed the threshold, the semantic analysis is carried out on the user question, and the answer related to the semantic result is searched from the knowledge base or the Internet, so that the response is timely and accurate, and the accuracy of the semantic analysis can be improved by adjusting the spoken language expression, so that the accuracy of the intelligent response is improved, and the reliability of the intelligent interaction is also improved.
Referring to fig. 2, the semantic parsing of the user question in S170 to obtain a semantic result includes the following sub-steps:
s171: and carrying out semantic analysis on the user question to obtain a plurality of semantic results.
The method specifically comprises the steps of segmenting words of user questions according to at least one of positions of users, service scenes of the users and language habits of the users, selecting at least one keyword in the user questions from the segmentation results or selecting at least one keyword, and forming a plurality of semantic results of the user questions by using different semantic annotations of the at least one keyword.
Since the language expression of users in different places is different, the word segmentation for sentences is also different. The language habits of different users are different, the intelligent interaction equipment can collect historical input information of the users, and establish a word segmentation model of the users aiming at the feedback of semantic results obtained after word segmentation of the users every time, the word segmentation model records the word segmentation mode of the users, and then word segmentation is carried out on the problems of the current users according to the word segmentation model. For example, if the current service scene is a game service scene, the word segmentation is "who is the lying bottom" of the current scene setting noun is not split, and the word segmentation is "who is the lying bottom", "regular"; if the current service scenario is a general service question and answer service scenario, the word is "who", "yes", "bedridden", "regular". Therefore, the intelligent interaction equipment can perform word segmentation on the user problem according to at least one of the position of the user, the service scene and the language habit of the user. If the participles are divided according to the position of the user, the service scene and the language habits of the user, weights can be set for the position of the user, the service scene and the language habits of the user, and the participle with the highest weight is selected for different participles obtained according to the position of the user, the service scene and the language habits of the user. For example, the word segmentation obtained according to the location of the user is "who", "yes", "lying", and the word segmentation obtained according to the service scenario is "who is lying", then the word segmentation obtained according to the service scenario with high weight is selected as "who is lying", or the word segmentation obtained according to the location of the user and the language habit of the user is "who", "yes", and "lying", and the word segmentation obtained according to the service scenario is "who is lying", then the word segmentation obtained according to the location of the user and the language habit of the user is "who is lying", and then the word segmentation obtained according to the location of the user and the language habit of the user is "who", "yes", and "lying".
Specifically, the word segmentation method may be, for example, "maximum probability method word segmentation", "maximum matching word segmentation", "dictionary matching algorithm", or the like. The dictionary matching algorithm includes at least one of a forward match, a reverse match, a bi-directional match, a maximum match, and a minimum match. Further, after word segmentation, ontology instantiation can be performed on the obtained words so as to identify information such as objects, properties, categories and the like of the words. The ontology is a specific detailed description of the concept, a description method of the real world, or a formal expression of a certain concept and its relationship in a specific field. After local instantiation, the plurality of words can obtain the attributes of the ontology, and preparation is made for semantic annotation analysis.
In addition, before word segmentation, denoising and modular structuring processing can be performed on the obtained user problems.
S172: the current semantic scene type is determined from the detected scene information.
The scene information comprises at least one of an application system or an application program used by a user, current operation information of the user in the application system or the application program, historical operation information of the user in the application system or the application program, context information, user identity information and collected current environment information. The application system or application used by the user is the application system or application currently running on the intelligent interactive device, for example, a travel-related application is running, and thus can be determined as a travel-related semantic scene type. The current operational information of the user at the application system or application is, for example, searching for a piece of athletic equipment in a shopping application, from which a semantic scene type associated with the piece of athletic equipment may be determined. The context information is the natural language input by the user history, and the current semantic scene can be obtained by analyzing the context information. The user identity information is professional information of the user, such as students, gourmets, construction engineers, athletes and the like, and the semantic scene can be automatically determined to be related to the identity according to the identity information of the user. The collected current environment information can include environment noise, a current position, a current time and the like, the environment where the user is located can be determined according to the information, and then a semantic scene determined to be related is obtained, for example, the environmental noise is analyzed to obtain disordered vehicle sounds, and the current time is in a peak period of working and working, so that the current semantic scene can be determined to be a congested road.
In an embodiment, when the obtained user question includes voice information, the detected scene information may further include a type of the input voice information, and the type of the voice information includes a normal speaking type and a singing type. The intelligent interactive device may determine the type of the voice information by detecting the intonation of the voice information, and select a semantic scene matching the type, for example, if the type is a singing type, a semantic scene related to a song is determined.
The intelligent interaction device can establish a classification model for each scene information so as to preset the corresponding semantic scene type of each scene under different conditions. After the scene information is detected, classifying each kind of scene information by using the classification model to obtain a corresponding preset semantic scene type, and determining the current semantic scene type.
Wherein, the intelligent interactive device may set different weights for each kind of scene information, this S162 includes: classifying each detected scene information to obtain a preset semantic scene type corresponding to each scene information, and selecting one of the obtained preset semantic scene types as a current semantic scene type according to the weight of each detected scene information. For example, when the detected scene information includes more than two types, and the intelligent interaction device obtains a plurality of preset semantic scene types according to the preset semantic scene types corresponding to each type of scene information, the preset semantic scene type with the highest weight corresponding to the scene information can be selected as the current semantic scene type; or selecting more than two preset semantic scene types with the highest weight as undetermined semantic scene types, dividing the rest preset semantic scene types into the undetermined semantic scene types according to the semantic scene similarity, adding the weights corresponding to all the preset semantic scene types divided into the same undetermined semantic scene type to be used as the total weight of the undetermined semantic scene types, and selecting the undetermined semantic scene type with the highest total weight as the current semantic scene type.
S173: and acquiring the determined feature information of the semantic scene type, and selecting the semantic result with the highest matching degree with the acquired feature information from the plurality of semantic results.
Specifically, the feature information of the semantic scene type includes at least one of a hot word, a common word, and a related word in the semantic scene type. For example, if the semantic scene type is sports, the intelligent interactive device collects hot words, common words, and associated words, such as "female jackpot game", "swimming", etc., related to sports on the network in the last period of time (e.g., one month). The intelligent interaction device can collect hot words with the use frequency higher than the set frequency and associated words matched with the hot words with the occurrence frequency higher than the set value from a set social platform, such as a microblog, a sticking bar and the like, and store the hot words and the associated words in a local database.
The intelligent interaction device obtains the feature information associated with the semantic scene type determined in S172 from the local database, and selects a semantic result whose semantic is most similar to the feature information from the plurality of semantic results obtained in S171.
In the embodiment, the current semantic scene type is determined through the detected scene information, and the semantic result of the user problem is determined through the feature information of the current semantic scene type, so that corresponding operation is realized according to the determined semantic result.
Referring to fig. 3, fig. 3 is a partial flowchart of another embodiment of the intelligent interaction method of the present application. The above S110 includes the following substeps:
s111: and receiving the voice information and the first text information input by the user, and carrying out voice recognition on the voice information to obtain second text information.
S112: and combining the first text information group and the second text information group into third text information according to the input sequence, wherein the third text information is used as the user question.
The embodiment adopts a mode that voice information input by a user and text information input by the user form a complete sentence according to the input sequence. For example, the user inputs text information "in the Shuihu pass", then inputs "Li Kui" by voice, and then inputs "introduction" by text, and obtains the text information "introduction of Li Kui in the Shuihu pass" by voice recognition and text combination. Therefore, the mode of matching the text and the voice input is adopted, even if the user encounters a word which is difficult to input the text, the user can select the voice input, and on the contrary, the user can also input the text for the word which cannot be read, so that the information input of the user is greatly facilitated. Further, the intelligent interaction device may obtain the result obtained through speech recognition in combination with the word sense of the first text information of the text input, for example, obtain two similar text results through speech recognition, and may select a reasonable text result in combination with the word sense of the first text information of the text input.
In another implementation, the intelligent interactive device may adopt semantic information and text information input by the user as complete sentences, and obtain a final semantic result by comparing the semantics of the two complete sentences. Specifically, the intelligent interaction device obtains first text information input by a user text, and obtains independent second text information through voice recognition. The intelligent interaction device executes subsequent steps on the first text information and the second text information until S170 is executed, a plurality of first semantic results corresponding to the first text information and a plurality of second semantic results corresponding to the second text information are obtained through semantic analysis, a first semantic result with the matching degree of the second semantic result exceeding a set threshold value is obtained from the first semantic results or a second semantic result with the matching degree of the first semantic result exceeding the set threshold value is obtained from the second semantic results, and the selected first semantic result or the selected second semantic result is the obtained plurality of semantic results.
Referring to fig. 4, fig. 4 is a schematic structural diagram of an embodiment of an intelligent interactive device according to the present application. In this embodiment, the intelligent interaction device 40 may specifically be a terminal or a server such as a computer and a mobile phone, or any device with processing capability such as a robot. The intelligent interaction device 40 comprises a memory 41, a processor 42, an input means 43 and an output means 44. Wherein, each component of the intelligent interactive device 40 can be coupled together through a bus, or the processor 42 of the intelligent interactive device 40 is connected with other components one by one.
The input device 43 is used for generating information in response to a user input operation or receiving information input by a user from another input device. For example, the input device 43 is a keyboard for generating corresponding text information in response to pressing of the keyboard by a user, the input device 43 is a touch screen for generating corresponding text information in response to touching by the user, the input device 43 is a microphone for generating corresponding voice information in response to voice of the user, the input device 43 is a receiver for receiving text, voice information and the like sent by other devices.
The output device 44 is used to feed information back to the user or other device user. Such as a display screen, player or transmitter etc.
The memory 41 stores a knowledge base storing questions and corresponding answers.
The memory 41 is also used for storing computer instructions executed by the processor 42 and data of the processor 42 in the process, wherein the memory 41 comprises a nonvolatile storage part for storing the computer instructions.
In this embodiment, processor 42, by invoking computer instructions stored by memory 41, is configured to:
receiving user questions obtained through the input device 43;
calculating the similarity between the user question and a pre-stored question in a knowledge base stored in a memory 41;
if the pre-stored questions with the similarity exceeding the threshold value with the user question exist in the knowledge base, outputting answers corresponding to the pre-stored questions with the similarity exceeding the threshold value in the knowledge base through an output device 44 according to the sequence of the similarity; adjusting the calculation mode of the similarity according to the selection record of the user on the output answer, so that the similarity of the pre-stored questions corresponding to the user question and the answer selected by the user is the highest;
if the knowledge base does not have the pre-stored problem of which the similarity with the user problem exceeds a threshold value, performing statement adjustment on the spoken language expression of the user problem, and recalculating the similarity between the adjusted user problem and the pre-stored problem in the knowledge base; if the similarity does not exceed the threshold, performing semantic analysis on the user question to obtain a semantic result, and searching answers related to the semantic result from the knowledge base or on the internet through the output device 44.
Optionally, the processor 42 performs the calculating of the similarity between the user question and the pre-stored question in the knowledge base, including: and determining the multi-dimensional similarity sequencing of the user questions and the pre-stored questions in the knowledge base based on different keywords of the user questions, and synthesizing the similarity sequencing of each dimension to obtain the similarity of the user questions and the pre-stored questions in the knowledge base.
Optionally, the processor 42 executes the search for the answer related to the semantic result through the output device 44, including: sending the user question to a user in the field of the question or a user who has answered other questions containing the keywords of the user question in an instant messaging manner through an output device 44 and requiring the user to answer in a limited time; the user response is received via the input device 43 in an instant messaging manner, and the received user response is returned via the output device 44.
Optionally, the processor 42 performs the semantic parsing on the user question to obtain a semantic result, including: performing semantic analysis on the user problem to obtain a plurality of semantic results; determining a current semantic scene type according to the detected scene information, wherein the scene information comprises at least one of an application system or an application program used by a user, current operation information of the user in the application system or the application program, historical operation information of the user in the application system or the application program, context information, user identity information and collected current environment information; and acquiring the determined feature information of the semantic scene type, and selecting the semantic result with the highest matching degree with the acquired feature information from the plurality of semantic results.
Further, the feature information of the semantic scene type includes at least one of a hot word, a common word, and a relevant word in the semantic scene type.
Further, processor 42 performs the semantic parsing on the user question to obtain a plurality of semantic results, including: segmenting the text information according to at least one of the position of the user, the service scene and the language habit of the user, and selecting at least one keyword in the text information from the segmentation result; and forming a plurality of semantic results of the text information by using different semantic annotations of the at least one keyword.
Optionally, the processor 42 performs the receiving of the user question obtained through the input device 43 includes: receiving voice information input by a user and first text information obtained through the input device 43, and performing voice recognition on the voice information to obtain second text information; and combining the first text information group and the second text information group into third text information according to the input sequence, wherein the third text information is used as the user question.
Optionally, the processor 42, after performing the searching for the answer related to the semantic result from the knowledge base or the internet, is further configured to: storing the user question and the searched related answer in a knowledge base;
optionally, the processor 42 performs the calculation method of the similarity after the adjustment according to the selection record of the output answer by the user, including: determining the expression habit of the user question according to the selection record of the user to the output answer; and adjusting the calculation mode of the similarity according to the determined expression habit.
In another embodiment, the processor 42 of the intelligent interactive device 40 may be used to perform the steps of the above-described example method of implementation.
Referring to fig. 5, the present application further provides an embodiment of a non-volatile storage medium, the non-volatile storage medium 50 stores a computer program 51 that can be executed by a processor, and the computer program 51 is used for executing the method in the foregoing embodiment. Specifically, the storage medium may be specifically the memory 41 shown in fig. 4.
In the above scheme, the intelligent interaction device outputs the answer corresponding to the pre-stored question by calculating the similarity between the user question and the pre-stored question, and outputs the highest similarity between the user question and the pre-stored question corresponding to the answer selected by the user according to the calculation mode of the similarity after the user selects and records the output answer, so that the calculation of the subsequent similarity can be ensured to be matched with the expression habit of the user, the accuracy of intelligent response is improved, and the reliability of intelligent interaction is improved; when the pre-stored problem with the similarity exceeding the threshold value is not found, sentence adjustment is carried out on the spoken expression of the user problem, and the similarity between the adjusted user problem and the pre-stored problem in the knowledge base is recalculated; if the similarity does not exceed the threshold, the semantic analysis is carried out on the user question, and the answer related to the semantic result is searched from the knowledge base or the Internet, so that the response is timely and accurate, and the accuracy of the semantic analysis can be improved by adjusting the spoken language expression, so that the accuracy of the intelligent response is improved, and the reliability of the intelligent interaction is also improved.
In the description above, for purposes of explanation and not limitation, specific details are set forth such as particular system structures, interfaces, techniques, etc. in order to provide a thorough understanding of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
Claims (7)
1. An intelligent interaction method, comprising:
receiving a user question;
calculating the similarity between the user question and a pre-stored question in a knowledge base;
if the pre-stored questions with the similarity exceeding the threshold value with the user question exist in the knowledge base, outputting answers corresponding to the pre-stored questions with the similarity exceeding the threshold value in the knowledge base according to the sequence of the similarity; adjusting the calculation mode of the similarity according to the selection record of the user on the output answer, so that the similarity of the pre-stored questions corresponding to the user question and the answer selected by the user is the highest;
if the knowledge base does not have the pre-stored problem of which the similarity with the user problem exceeds a threshold value, performing statement adjustment on the spoken language expression of the user problem, and recalculating the similarity between the adjusted user problem and the pre-stored problem in the knowledge base; if the similarity does not exceed the threshold value, performing semantic analysis on the user question to obtain a semantic result, and searching answers related to the semantic result from the knowledge base or the Internet;
the semantic parsing of the user problem to obtain a semantic result comprises the following steps:
performing semantic analysis on the user problem to obtain a plurality of semantic results;
determining a current semantic scene type according to the detected scene information, wherein the scene information comprises at least one of an application system or an application program used by a user, current operation information of the user in the application system or the application program, historical operation information of the user in the application system or the application program, context information, user identity information and collected current environment information;
acquiring the determined feature information of the semantic scene type, and selecting a semantic result with the highest matching degree with the acquired feature information from the plurality of semantic results, wherein the feature information of the semantic scene type comprises at least one of a hot word, a common word and a relevant word under the semantic scene type;
the semantic parsing of the user question to obtain a plurality of semantic results comprises:
performing word segmentation on the user question according to at least one of the position of the user, the service scene and the language habit of the user, and selecting at least one keyword in the user question from the word segmentation result;
and forming a plurality of semantic results for obtaining the user question by using different semantic annotations of the at least one keyword.
2. The method of claim 1, wherein the calculating the similarity between the user question and the pre-stored question in the knowledge base comprises:
and determining the multi-dimensional similarity sequencing of the user questions and the pre-stored questions in the knowledge base based on different keywords of the user questions, and synthesizing the similarity sequencing of each dimension to obtain the similarity of the user questions and the pre-stored questions in the knowledge base.
3. The method of claim 1, wherein the searching for the answer related to the semantic result from the internet comprises:
sending the user question to a user in the field of the question or a user who has answered other questions containing the keywords of the user question through instant messaging, and requiring the user to answer in a limited time;
and returning the user response obtained through the instant messaging.
4. The method of claim 1, wherein the receiving a user question comprises:
receiving voice information and first text information input by the user, and performing voice recognition on the voice information to obtain second text information;
and combining the first text information group and the second text information group into third text information according to the input sequence, wherein the third text information is used as the user question.
5. The method of claim 1, further comprising, after said searching answers related to said semantic results from said knowledge base or internet, the steps of:
storing the user question and the searched related answer in a knowledge base;
the calculation method for adjusting the similarity according to the selection record of the user for the output answer comprises the following steps:
determining the expression habit of the user question according to the selection record of the user to the output answer; and adjusting the calculation mode of the similarity according to the determined expression habit.
6. An intelligent interaction device, comprising a memory and a processor connected to each other;
the processor is configured to perform the method of any one of claims 1 to 5.
7. A non-volatile storage medium, characterized in that a computer program is stored for execution by a processor for performing the method of any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710815146.5A CN107609101B (en) | 2017-09-11 | 2017-09-11 | Intelligent interaction method, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710815146.5A CN107609101B (en) | 2017-09-11 | 2017-09-11 | Intelligent interaction method, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107609101A CN107609101A (en) | 2018-01-19 |
| CN107609101B true CN107609101B (en) | 2020-10-27 |
Family
ID=61062446
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710815146.5A Active CN107609101B (en) | 2017-09-11 | 2017-09-11 | Intelligent interaction method, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107609101B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11841867B2 (en) | 2019-08-09 | 2023-12-12 | International Business Machines Corporation | Query relaxation using external domain knowledge for query answering |
Families Citing this family (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108287648A (en) * | 2018-01-31 | 2018-07-17 | 泰康保险集团股份有限公司 | Feedback management method, apparatus, electronic equipment and the computer storage media of software |
| CN108459733A (en) * | 2018-02-06 | 2018-08-28 | 广州阿里巴巴文学信息技术有限公司 | auxiliary input method, device, computing device and storage medium |
| CN108388558B (en) * | 2018-02-07 | 2022-04-19 | 平安普惠企业管理有限公司 | Question matching method and device, customer service robot and storage medium |
| CN108491433B (en) * | 2018-02-09 | 2022-05-03 | 平安科技（深圳）有限公司 | Chat response method, electronic device and storage medium |
| CN110164427A (en) * | 2018-02-13 | 2019-08-23 | 阿里巴巴集团控股有限公司 | Voice interactive method, device, equipment and storage medium |
| CN110334177B (en) * | 2018-03-15 | 2023-05-30 | 阿里巴巴集团控股有限公司 | Semantic similarity model training and semantic similarity recognition methods and devices and electronic equipment |
| CN110413985B (en) * | 2018-04-27 | 2022-09-16 | 北京海马轻帆娱乐科技有限公司 | Related text segment searching method and device |
| CN109697244A (en) * | 2018-11-01 | 2019-04-30 | 百度在线网络技术（北京）有限公司 | Information processing method, device and storage medium |
| CN109545202B (en) * | 2018-11-08 | 2021-05-11 | 广东小天才科技有限公司 | Method and system for adjusting corpus with semantic logic confusion |
| CN109684632B (en) * | 2018-12-12 | 2023-04-21 | 达闼机器人股份有限公司 | Natural semantic understanding method, device and computing equipment |
| CN109635091A (en) * | 2018-12-14 | 2019-04-16 | 上海钛米机器人科技有限公司 | A kind of method for recognizing semantics, device, terminal device and storage medium |
| CN109739962A (en) * | 2018-12-26 | 2019-05-10 | 广州灵聚信息科技有限公司 | A kind of control method and device of Chat mode |
| CN110020429B (en) * | 2019-02-27 | 2023-05-23 | 阿波罗智联(北京)科技有限公司 | Semantic recognition method and device |
| CN110211576B (en) * | 2019-04-28 | 2021-07-30 | 北京蓦然认知科技有限公司 | Voice recognition method, device and system |
| CN110321416A (en) * | 2019-05-23 | 2019-10-11 | 深圳壹账通智能科技有限公司 | Intelligent answer method, apparatus, computer equipment and storage medium based on AIML |
| CN110188248A (en) * | 2019-05-28 | 2019-08-30 | 新华网股份有限公司 | Data processing method, device and electronic equipment based on news question and answer interactive system |
| CN110189047A (en) * | 2019-06-04 | 2019-08-30 | 甘肃万华金慧科技股份有限公司 | A kind of station Admin Administration system |
| WO2021000403A1 (en) * | 2019-07-03 | 2021-01-07 | 平安科技（深圳）有限公司 | Voice matching method for intelligent dialogue system, electronic device and computer device |
| CN110647614B (en) * | 2019-08-01 | 2023-05-23 | 平安科技（深圳）有限公司 | Intelligent question-answering method, device, medium and electronic equipment |
| CN110517688A (en) * | 2019-08-20 | 2019-11-29 | 合肥凌极西雅电子科技有限公司 | A kind of voice association prompt system |
| CN110931012A (en) * | 2019-10-12 | 2020-03-27 | 深圳壹账通智能科技有限公司 | Reply message generation method and device, computer equipment and storage medium |
| CN111222043A (en) * | 2019-12-31 | 2020-06-02 | 联想(北京)有限公司 | Information processing method and system and electronic equipment |
| CN113111155B (en) * | 2020-01-10 | 2024-04-19 | 阿里巴巴集团控股有限公司 | Information display method, device, equipment and storage medium |
| CN113220848A (en) * | 2020-01-21 | 2021-08-06 | 华为技术有限公司 | Automatic question answering method and device for man-machine interaction and intelligent equipment |
| CN111583919B (en) * | 2020-04-15 | 2023-10-13 | 北京小米松果电子有限公司 | Information processing method, device and storage medium |
| CN111966840A (en) * | 2020-08-18 | 2020-11-20 | 北京猿力未来科技有限公司 | Man-machine interaction management method and management system for language teaching |
| CN112115244B (en) * | 2020-08-21 | 2024-05-03 | 深圳市欢太科技有限公司 | Dialogue interaction method and device, storage medium and electronic equipment |
| CN113743124B (en) * | 2021-08-25 | 2024-03-29 | 南京星云数字技术有限公司 | Intelligent question-answering exception processing method and device and electronic equipment |
| CN114518553B (en) * | 2022-04-14 | 2022-07-19 | 中国科学院精密测量科学与技术创新研究院 | Broadband frequency conversion pseudo-two-dimensional spectrum NMR method for accurately measuring field intensity of electromagnet |
| CN116340481B (en) * | 2023-02-27 | 2024-05-10 | 华院计算技术(上海)股份有限公司 | Method and device for automatically replying to question, computer readable storage medium and terminal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199810A (en) * | 2014-08-29 | 2014-12-10 | 科大讯飞股份有限公司 | Intelligent service method and system based on natural language interaction |
| CN104679910A (en) * | 2015-03-25 | 2015-06-03 | 北京智齿博创科技有限公司 | Intelligent answering method and system |
| US9613025B2 (en) * | 2014-11-19 | 2017-04-04 | Electronics And Telecommunications Research Institute | Natural language question answering system and method, and paraphrase module |
-
2017
- 2017-09-11 CN CN201710815146.5A patent/CN107609101B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104199810A (en) * | 2014-08-29 | 2014-12-10 | 科大讯飞股份有限公司 | Intelligent service method and system based on natural language interaction |
| US9613025B2 (en) * | 2014-11-19 | 2017-04-04 | Electronics And Telecommunications Research Institute | Natural language question answering system and method, and paraphrase module |
| CN104679910A (en) * | 2015-03-25 | 2015-06-03 | 北京智齿博创科技有限公司 | Intelligent answering method and system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11841867B2 (en) | 2019-08-09 | 2023-12-12 | International Business Machines Corporation | Query relaxation using external domain knowledge for query answering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107609101A (en) | 2018-01-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107609101B (en) | Intelligent interaction method, equipment and storage medium | |
| CN107797984B (en) | Intelligent interaction method, equipment and storage medium | |
| CN107818781B (en) | Intelligent interaction method, equipment and storage medium | |
| CN107832286B (en) | Intelligent interaction method, equipment and storage medium | |
| CN108829757B (en) | Intelligent service method, server and storage medium for chat robot | |
| CN110121706B (en) | Providing responses in a conversation | |
| CN108647205B (en) | Fine-grained emotion analysis model construction method and device and readable storage medium | |
| WO2020108608A1 (en) | Search result processing method, device, terminal, electronic device, and storage medium | |
| CN104836720B (en) | Method and device for information recommendation in interactive communication | |
| KR102288249B1 (en) | Information processing method, terminal, and computer storage medium | |
| CN110795542B (en) | Dialogue method, related device and equipment | |
| CN111602147A (en) | Machine learning model based on non-local neural network | |
| CN107832432A (en) | A kind of search result ordering method, device, server and storage medium | |
| WO2017097061A1 (en) | Smart response method and apparatus | |
| WO2018040069A1 (en) | Information recommendation system and method | |
| CN108304375A (en) | A kind of information identifying method and its equipment, storage medium, terminal | |
| CN109918555B (en) | Method, apparatus, device and medium for providing search suggestions | |
| CN108920649B (en) | Information recommendation method, device, equipment and medium | |
| CN111090771B (en) | Song searching method, device and computer storage medium | |
| CN110147494B (en) | Information searching method and device, storage medium and electronic equipment | |
| CN109615009B (en) | Learning content recommendation method and electronic equipment | |
| JP7096172B2 (en) | Devices, programs and methods for generating dialogue scenarios, including utterances according to character. | |
| JPWO2016178337A1 (en) | Information processing apparatus, information processing method, and computer program | |
| CN109977292B (en) | Search method, search device, computing equipment and computer-readable storage medium | |
| CN112528046B (en) | New knowledge graph construction method and device and information retrieval method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |