CN110427626A - The extracting method and device of keyword - Google Patents
The extracting method and device of keyword Download PDFInfo
- Publication number
- CN110427626A CN110427626A CN201910703459.0A CN201910703459A CN110427626A CN 110427626 A CN110427626 A CN 110427626A CN 201910703459 A CN201910703459 A CN 201910703459A CN 110427626 A CN110427626 A CN 110427626A
- Authority
- CN
- China
- Prior art keywords
- word
- value
- text
- weighted value
- keyword
- 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.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of extracting method of keyword and devices.Specifically, this method comprises: the text to input carries out text retrieval conference TREC, with corresponding first weighted value of content type each in the determination text；Semantic analysis is carried out to each word in the text, with corresponding second weighted value of each word of determination；The term frequencies TF value of each word is adjusted according to first weighted value and second weighted value, and the TF value by adjusting after calculates the third weighted value of each word；According to the third weighted value, extracts in the word and specify word as the keyword for being retrieved.Through the invention, it solves TF-IDF to calculate dependent on the relevant document of multiple contents, the word weight of single text can not be calculated, and TF-IDF method discrete data lower for the degree of association shows poor problem, reach the precision effect for improving and extracting to key word information.
Description
Technical field
The present invention relates to the communications fields, in particular to the extracting method and device of a kind of keyword.
Background technique
The current most common searching system is all based on what keyword was realized, and the extraction of keyword, nearly all uses word
Frequently the calculation method of (term frequency, TF) and anti-document frequency (inverse document frequency, IDF).
But TF-IDF is calculated and is depended on the relevant document of multiple contents, can not calculate the word weight of single text, and the side TF-IDF
Method discrete data performance lower for the degree of association is poor.
Summary of the invention
The embodiment of the invention provides a kind of extracting method of keyword and devices, at least to solve TF- in the related technology
IDF, which is calculated, depends on the relevant document of multiple contents, can not calculate the word weight of single text, and TF-IDF method is for closing
The lower discrete data of connection degree shows poor problem.
According to one embodiment of present invention, provide a kind of extracting method of keyword, comprising: to the text of input into
Row text retrieval conference TREC, with corresponding first weighted value of content type each in the determination text；To each in the text
A word carries out semantic analysis, with corresponding second weighted value of each word of determination；According to first weighted value and described
Two weighted values are adjusted the term frequencies TF value of each word, and the TF value by adjusting after calculates each institute
The third weighted value of predicate language；According to the third weighted value, extracts in the word and word is specified to be used as being retrieved
Keyword.
Optionally, before carrying out semantic analysis to the word in the text, the method also includes: according to default rule
Word segmentation processing then is carried out to the text, and, it is determined according to the relevance between each word after participle each
The part of speech of a word.
Optionally, semantic analysis is carried out to each word in the text, to determine corresponding second power of specified word
Weight values, comprising: each word is ranked up according to preset part of speech priority rule；According to the part of speech priority
Sequence assigns corresponding second weighted value to each word.
Optionally, according to first weighted value and second weighted value to the term frequencies TF value of each word
It is adjusted, further includes: obtain the TF value of each word；By the TF value and first weighted value and described the
Two weighted values are multiplied, with the determination TF value adjusted.
Optionally, the TF value by adjusting after calculates the third weighted value of each word, comprising: obtains each
The IDF value of the word；It is determined according to the TF value of each word adjusted and the IDF value of each word
The third weighted value.
Optionally, it according to the third weighted value, extracts in the word and specifies word as the pass for being retrieved
Keyword, the method also includes: it removes third weighted value described in the word and is less than the word of default weight threshold as institute
State specified word.
Optionally, the content type includes at least one of: drawing in the text according to pre-set text format
Point content of text type, the location type of paragraph in the text, the location type of sentence in the text.
According to another embodiment of the invention, a kind of extraction element of keyword is provided, comprising: first determines mould
Block, for carrying out text retrieval conference TREC to the text of input, with corresponding first power of content type each in the determination text
Weight values；Second determining module, it is corresponding with each word of determination for carrying out semantic analysis to each word in the text
Second weighted value；Module is adjusted, for the word according to first weighted value and second weighted value to each word
Speech frequency rate TF value is adjusted, and the TF value by adjusting after calculates the third weighted value of each word；Extract mould
Block specifies word as the keyword for being retrieved for according to the third weighted value, extracting in the word.
According to still another embodiment of the invention, a kind of storage medium is additionally provided, meter is stored in the storage medium
Calculation machine program, wherein the computer program is arranged to execute the step in any of the above-described embodiment of the method when operation.
According to still another embodiment of the invention, a kind of electronic device, including memory and processor are additionally provided, it is described
Computer program is stored in memory, the processor is arranged to run the computer program to execute any of the above-described
Step in embodiment of the method.
Through the invention, the TF of word is carried out using the results of structural analysis of text and the semantic analysis result of word
Adjustment, therefore, can solve solution, TF-IDF is calculated dependent on the relevant document of multiple contents in the related technology, can not calculate list
The word weight of one text, and TF-IDF method discrete data lower for the degree of association shows poor problem, reaches
Improve the precision effect extracted to key word information.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of the extraction of keyword according to an embodiment of the present invention；
Fig. 2 is a kind of schematic diagram of test text according to an embodiment of the present invention；
Fig. 3 is a kind of result figure for extracting result according to an embodiment of the present invention；
Fig. 4 is a kind of structural block diagram of the extracting method device of keyword according to an embodiment of the present invention.
Specific embodiment
Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and in combination with Examples.It should be noted that not conflicting
In the case of, the features in the embodiments and the embodiments of the present application can be combined with each other.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.
Embodiment 1
A kind of extracting method for running on keyword is provided in the present embodiment, and Fig. 1 is according to an embodiment of the present invention
The flow chart of the extraction of keyword, as shown in Figure 1, the process includes the following steps:
Step S102 carries out text retrieval conference TREC to the text of input, with each content type pair in the determination text
The first weighted value answered；
Step S104 carries out semantic analysis to each word in the text, with each word of determination corresponding second
Weighted value；
Step S106, according to first weighted value and second weighted value to the term frequencies TF of each word
Value is adjusted, and the TF value by adjusting after calculates the third weighted value of each word；
Step S108 is extracted in the word and is specified word as being retrieved according to the third weighted value
Keyword.
Optionally, the content type includes at least one of: drawing in the text according to pre-set text format
Point content of text type, the location type of paragraph in the text, the location type of sentence in the text.
Specifically, in the text according to pre-set text format divide content of text type refer in text it is each in
Hold part.Such as the specification of a patent application, abstract of description, abstract of description attached drawing, claims,
Specification, Figure of description can be used as a kind of division mode of content of text type of content part.And for specification and
Speech, the division mode of the content of text type of the specification can be said according to technical field, background technique, summary of the invention, attached drawing
Bright, specific embodiment is divided.
Specifically, the location type of paragraph refers to the position of paragraph content of text in the text in text.Such as, if
In technical field, if in summary of the invention, if in the description of the drawings.Whether paragraph is first paragraph simultaneously, still most
Paragraph afterwards, or the paragraph of some intermediate position.
Specifically, the location type of sentence is similar with the location type of paragraph in text in text, refer to sentence in section
It falls or the position of content of text in the text.Such as, if in technical field, if in summary of the invention, if attached
In figure explanation.Simultaneously whether some paragraph first sentence, tail sentence or middle section.
Above description is illustrative examples, and the manifestation mode of any content type based on above-mentioned thinking is in this implementation
Within the protection scope of example.
Specifically, being with patent application when corresponding first weighted value of each content type in determining the text
, content importance of the content obviously than other four parts in specification in specific embodiment is high, therefore, for
Paragraph, sentence in the specific embodiment, specific embodiment can assign weight more higher than other content type.And having
In body embodiment, often the content in first section or preceding several paragraphs is most important.Therefore, for first section or former
A paragraph will assign weight more higher than other paragraphs.And in each paragraph, often first section or endpiece usually provide knot
By the sentence of property, therefore, the weight of head and the tail section will assign weight more higher than sentence in other paragraphs.
Optionally, before carrying out semantic analysis to the word in the text, the method also includes: according to default rule
Word segmentation processing then is carried out to the text, and, it is determined according to the relevance between each word after participle each
The part of speech of a word.
Optionally, semantic analysis is carried out to each word in the text, to determine corresponding second power of specified word
Weight values, comprising: each word is ranked up according to preset part of speech priority rule；According to the part of speech priority
Sequence assigns corresponding second weighted value to each word.
Specifically, on the one hand semantic analysis can filter out the principal entities of article discussion according to semanteme, on the other hand,
The unwanted contributions in sentence can be removed.For example, " Xiao Ming is a Chinese ", can propose subject " Xiao Ming " and predicative " in
Compatriots ".Then a higher weight, other words are assigned for critical entities such as subject, the objects of core phrase and sentence
Weight of converging is directly disposed as 1.Another example is that in view of sometimes quantifier, adjective are also more crucial part.Cause
This, can to quantifier, adjectival weight assign it is lower than critical entities such as core phrase and the subject of sentence, objects, but than it
The high weight of his word can assign higher value.
Optionally, according to first weighted value and second weighted value to the term frequencies TF value of each word
It is adjusted, further includes: obtain the TF value of each word；By the TF value and first weighted value and described the
Two weighted values are multiplied, with the determination TF value adjusted.
Optionally, the TF value by adjusting after calculates the third weighted value of each word, comprising: obtains each
The IDF value of the word；It is determined according to the TF value of each word adjusted and the IDF value of each word
The third weighted value.
Optionally, it according to the third weighted value, extracts in the word and specifies word as the pass for being retrieved
Keyword, the method also includes: it removes third weighted value described in the word and is less than the word of default weight threshold as institute
State specified word.
It should be pointed out that the purpose of default weight threshold is to influence the knot retrieved below in order to avoid output result is excessive
Fruit, because excessive keyword is searched in knowledge mapping, it is possible to can because major key is excessive, return excessive information or
Person excessively can not return information because limiting.
In order to better understand the technical solution recorded in the present embodiment, following scene is additionally provided in the present embodiment
To better understand the scheme recorded in above-described embodiment.
Fig. 2 is a kind of schematic diagram of test text according to an embodiment of the present invention.As shown in Fig. 2,
Step 1: text retrieval conference TREC is carried out to the test text of input.To analyze, the 1st row is assigned, the 31st row is most
High weighted value assigns the 7th row, the weighted value (being equivalent to the content in the 7th row of removal and the 9th row) that the 9th row weighted value is 0,
However the sentence of other rows is then assigned lower than the 1st row, the weighted value of the 31st row.
Step 2: semantic analysis being carried out to each word in the text, is segmented, part of speech, goes to listen word, syntax
The processing such as analysis.Such as recorded in the 2nd row " many people really remember artificial intelligence, or because Shi Di in 2001
This literary Pierre's Burger instructs that film " artificial intelligence " " it can extract, subject " people ", " Glenn Stevens Pierre Burger ", it calls
Language " remembers ", " guidance ", object " artificial intelligence ".However for the subject of core, object " people ", " Glenn Stevens Pierre primary
Lattice ", " artificial intelligence " assign the highest weighted value greater than 1.Predicate " remembers " that " guidance " then assigns and be greater than 1 but be less than core
Subject, the corresponding weighted value of object data.Other words are then directly disposed as 1.
Step 3: according to first weighted value and second weighted value to the term frequencies TF value of each word
It is adjusted.Calculate the TF value of each word in every words.And multiplication is carried out according to the weighted value that step 1 and step 2 are got
Operation gets the TF value of each word.
Step 4: calculating the TF-IDF value of each word, (the case where being zero for IDF value, TF-IDF value is directly weighed using word
Weight).It after obtaining the TF-IDF value of each word, can directly use, or utilize sigmoid function, TF-IDF value is gone
Linearisation, and result is normalized.To get the corresponding weighted value of keyword of each word.
Step 5: each word weighted value corresponding with keyword is compared with preset threshold value, thus filter out as
Result shown in Fig. 3.Fig. 3 is a kind of result figure for extracting result according to an embodiment of the present invention.As shown in figure 3, final extract
It as a result is " artificial intelligence ", " mankind ", " robot ", " machine ", " law ".Wherein, the weight highest of law ", " robot "
Weight it is minimum.
Step 6: according to the output in Fig. 3 as a result, targetedly being retrieved.
Through the above description of the embodiments, those skilled in the art can be understood that according to above-mentioned implementation
The method of example can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but it is very much
In the case of the former be more preferably embodiment.Based on this understanding, technical solution of the present invention is substantially in other words to existing
The part that technology contributes can be embodied in the form of software products, which is stored in a storage
In medium (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, calculate
Machine, server or network equipment etc.) execute method described in each embodiment of the present invention.
Embodiment 2
Additionally provide a kind of device in the present embodiment, the device for realizing above-described embodiment and preferred embodiment,
The descriptions that have already been made will not be repeated.As used below, term " module " may be implemented predetermined function software and/or
The combination of hardware.Although device described in following embodiment is preferably realized with software, hardware or software and hard
The realization of the combination of part is also that may and be contemplated.
Fig. 4 is a kind of structural block diagram of the extracting method device of keyword according to an embodiment of the present invention, as shown in figure 4,
The device includes:
First determining module 42, for carrying out text retrieval conference TREC to the text of input, with each in the determination text
Corresponding first weighted value of content type；
Second determining module 44, for carrying out semantic analysis to each word in the text, with each word of determination
Corresponding second weighted value；
Module 46 is adjusted, for the word according to first weighted value and second weighted value to each word
Frequency TF value is adjusted, and the TF value by adjusting after calculates the third weighted value of each word；
Extraction module 48 specifies word to be used as carrying out for according to the third weighted value, extracting in the word
The keyword of retrieval.
It should be noted that above-mentioned modules can be realized by software or hardware, for the latter, Ke Yitong
Following manner realization is crossed, but not limited to this: above-mentioned module is respectively positioned in same processor；Alternatively, above-mentioned modules are with any
Combined form is located in different processors.
Embodiment 3
The embodiments of the present invention also provide a kind of storage medium, computer program is stored in the storage medium, wherein
The computer program is arranged to execute the step in any of the above-described embodiment of the method when operation.
Optionally, in the present embodiment, above-mentioned storage medium can be set to store by executing based on following steps
Calculation machine program:
S1 carries out text retrieval conference TREC to the text of input, with each content type in the determination text corresponding the
One weighted value；
S2 carries out semantic analysis to each word in the text, with corresponding second weighted value of each word of determination；
S3 carries out the term frequencies TF value of each word according to first weighted value and second weighted value
Adjustment, and the TF value by adjusting after calculates the third weighted value of each word.
Optionally, in the present embodiment, above-mentioned storage medium can include but is not limited to: USB flash disk, read-only memory (Read-
Only Memory, referred to as ROM), it is random access memory (Random Access Memory, referred to as RAM), mobile hard
The various media that can store computer program such as disk, magnetic or disk.
The embodiments of the present invention also provide a kind of electronic device, including memory and processor, stored in the memory
There is computer program, which is arranged to run computer program to execute the step in any of the above-described embodiment of the method
Suddenly.
Optionally, above-mentioned electronic device can also include transmission device and input-output equipment, wherein the transmission device
It is connected with above-mentioned processor, which connects with above-mentioned processor.
Optionally, in the present embodiment, above-mentioned processor can be set to execute following steps by computer program:
S1 carries out text retrieval conference TREC to the text of input, with each content type in the determination text corresponding the
One weighted value；
S2 carries out semantic analysis to each word in the text, with corresponding second weighted value of each word of determination；
S3 carries out the term frequencies TF value of each word according to first weighted value and second weighted value
Adjustment, and the TF value by adjusting after calculates the third weighted value of each word.
Optionally, the specific example in the present embodiment can be with reference to described in above-described embodiment and optional embodiment
Example, details are not described herein for the present embodiment.
Obviously, those skilled in the art should be understood that each module of the above invention or each step can be with general
Computing device realize that they can be concentrated on a single computing device, or be distributed in multiple computing devices and formed
Network on, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
It is performed by computing device in the storage device, and in some cases, it can be to be different from shown in sequence execution herein
Out or description the step of, perhaps they are fabricated to each integrated circuit modules or by them multiple modules or
Step is fabricated to single integrated circuit module to realize.In this way, the present invention is not limited to any specific hardware and softwares to combine.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.It is all within principle of the invention, it is made it is any modification, etc.
With replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of extracting method of keyword characterized by comprising
Text retrieval conference TREC is carried out to the text of input, with corresponding first weight of content type each in the determination text
Value；
Semantic analysis is carried out to each word in the text, with corresponding second weighted value of each word of determination；
The term frequencies TF value of each word is adjusted according to first weighted value and second weighted value, and
The TF value by adjusting after calculates the third weighted value of each word；
According to the third weighted value, extracts in the word and specify word as the keyword for being retrieved.
2. the method according to claim 1, wherein in the text word carry out semantic analysis it
Before, the method also includes:
Word segmentation processing is carried out to the text according to preset rules, and,
The part of speech of each word is determined according to the relevance between each word after participle.
3. according to the method described in claim 2, it is characterized in that, in the text each word carry out semantic analysis,
To determine specified corresponding second weighted value of word, comprising:
Each word is ranked up according to preset part of speech priority rule；
Corresponding second weighted value is assigned to each word according to the sequence of the part of speech priority.
4. the method according to claim 1, wherein according to first weighted value and second weighted value pair
The term frequencies TF value of each word is adjusted, further includes:
Obtain the TF value of each word；
The TF value is multiplied with first weighted value and second weighted value, with the determination TF value adjusted.
5. the method according to claim 1, wherein the TF value by adjusting after calculates each word
Third weighted value, comprising:
Obtain the IDF value of each word；
The third weight is determined according to the IDF value of the TF value of each word adjusted and each word
Value.
6. the method according to claim 1, wherein extracting the word middle finger according to the third weighted value
Determine word as the keyword for being retrieved, the method also includes:
It removes third weighted value described in the word and is less than the word of default weight threshold as the specified word.
7. the method according to claim 1, wherein the content type includes at least one of:
The content of text type divided in the text according to pre-set text format, the location type of paragraph, institute in the text
State the location type of sentence in text.
8. a kind of extraction element of keyword characterized by comprising
First determining module, for carrying out text retrieval conference TREC to the text of input, with each content class in the determination text
Corresponding first weighted value of type；
Second determining module, it is corresponding with each word of determination for carrying out semantic analysis to each word in the text
Second weighted value；
Module is adjusted, for the term frequencies TF according to first weighted value and second weighted value to each word
Value is adjusted, and the TF value by adjusting after calculates the third weighted value of each word；
Extraction module is extracted and specifies word as being retrieved in the word for according to the third weighted value
Keyword.
9. a kind of storage medium, which is characterized in that be stored with computer program in the storage medium, wherein the computer
Program is arranged to execute method described in any one of claim 1 to 7 when operation.
10. a kind of electronic device, including memory and processor, which is characterized in that be stored with computer journey in the memory
Sequence, the processor are arranged to run the computer program to execute side described in any one of claim 1 to 7
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910703459.0A CN110427626B (en) | 2019-07-31 | 2019-07-31 | Keyword extraction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910703459.0A CN110427626B (en) | 2019-07-31 | 2019-07-31 | Keyword extraction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110427626A true CN110427626A (en) | 2019-11-08 |
| CN110427626B CN110427626B (en) | 2022-12-09 |
Family
ID=68413496
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910703459.0A Active CN110427626B (en) | 2019-07-31 | 2019-07-31 | Keyword extraction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110427626B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931480A (en) * | 2020-07-03 | 2020-11-13 | 北京新联财通咨询有限公司 | Method and device for determining main content of text, storage medium and computer equipment |
| CN112417101A (en) * | 2020-11-23 | 2021-02-26 | 平安科技（深圳）有限公司 | Keyword extraction method and related device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130185308A1 (en) * | 2012-01-13 | 2013-07-18 | International Business Machines Corporation | System and method for extraction of off-topic part from conversation |
| CN103927302A (en) * | 2013-01-10 | 2014-07-16 | 阿里巴巴集团控股有限公司 | Text classification method and system |
| CN108920488A (en) * | 2018-05-14 | 2018-11-30 | 平安科技（深圳）有限公司 | The natural language processing method and device that multisystem combines |
-
2019
- 2019-07-31 CN CN201910703459.0A patent/CN110427626B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130185308A1 (en) * | 2012-01-13 | 2013-07-18 | International Business Machines Corporation | System and method for extraction of off-topic part from conversation |
| CN103927302A (en) * | 2013-01-10 | 2014-07-16 | 阿里巴巴集团控股有限公司 | Text classification method and system |
| CN108920488A (en) * | 2018-05-14 | 2018-11-30 | 平安科技（深圳）有限公司 | The natural language processing method and device that multisystem combines |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111931480A (en) * | 2020-07-03 | 2020-11-13 | 北京新联财通咨询有限公司 | Method and device for determining main content of text, storage medium and computer equipment |
| CN112417101A (en) * | 2020-11-23 | 2021-02-26 | 平安科技（深圳）有限公司 | Keyword extraction method and related device |
| CN112417101B (en) * | 2020-11-23 | 2023-08-18 | 平安科技（深圳）有限公司 | Keyword extraction method and related device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110427626B (en) | 2022-12-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102431549B1 (en) | Causality recognition device and computer program therefor | |
| US11531818B2 (en) | Device and method for machine reading comprehension question and answer | |
| KR102163549B1 (en) | Method and apparatus for determining retreat | |
| CN112988969B (en) | Method, apparatus, device and storage medium for text retrieval | |
| US7099819B2 (en) | Text information analysis apparatus and method | |
| CN109299280B (en) | Short text clustering analysis method and device and terminal equipment | |
| CN110874528B (en) | Text similarity obtaining method and device | |
| CN110297893A (en) | Natural language question-answering method, device, computer installation and storage medium | |
| WO2021159655A1 (en) | Data attribute filling method, apparatus and device, and computer-readable storage medium | |
| KR20200137924A (en) | Real-time keyword extraction method and device in text streaming environment | |
| CN111309916A (en) | Abstract extraction method and device, storage medium and electronic device | |
| US8806455B1 (en) | Systems and methods for text nuclearization | |
| CN110427626A (en) | The extracting method and device of keyword | |
| US9454568B2 (en) | Method, apparatus and computer storage medium for acquiring hot content | |
| CN111401039A (en) | Word retrieval method, device, equipment and storage medium based on binary mutual information | |
| CN117076650B (en) | Intelligent dialogue method, device, medium and equipment based on large language model | |
| CN113392305A (en) | Keyword extraction method and device, electronic equipment and computer storage medium | |
| CN110287284B (en) | Semantic matching method, device and equipment | |
| CN117113174A (en) | Model training method and device, storage medium and electronic equipment | |
| JP2017027106A (en) | Similarity calculation device, similarity retrieval device, and similarity calculation program | |
| CN112784046B (en) | Text clustering method, device, equipment and storage medium | |
| CN113704433A (en) | Man-machine conversation voice intention recognition method, device, equipment and storage medium | |
| CN113934842A (en) | Text clustering method and device and readable storage medium | |
| CN112579769A (en) | Keyword clustering method and device, storage medium and electronic equipment | |
| CN111209752A (en) | Chinese extraction integrated unsupervised abstract method based on auxiliary information |
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 |