A conference summary generation system based on natural language processing

Abstract

The present application relates to the technical field of conference summary generation, and discloses a conference summary generation system based on natural language processing, including: a speaker identification module, a text processing module, a first keyword generation module, a first summary generation module, a second keyword generation module, and a second summary generation module. The present application selects any one of the first summary or the second summary as the conference summary of the current meeting. The present application realizes the conversion of conference records in video or audio format through the speaker identification module and the text processing module, thereby simplifying the process of organizing conference records; realizes the function of extracting conference summaries according to conference themes through the first keyword generation module and the first summary generation module; realizes keyword optimization based on the speaker's rank through the second keyword generation module and the second summary generation module, thereby improving the quality of conference summaries and avoiding unnecessary waste of manpower and material resources.

Description

Technical Field

[0001] The present application relates to the technical field of conference summary generation, and specifically to a conference summary generation system based on natural language processing. Background Art

[0002] With the improvement of modern office technology, online meetings are becoming increasingly common, bringing with them meeting records saved in audio or video format. The delay of the communication network increases the difficulty of having a dedicated person to take meeting records, but finding someone to organize them after the meeting will increase additional manpower and material costs. Summary of the invention

[0003] The purpose of this application is to provide a conference summary generation system based on natural language processing to solve the technical problems raised in the above background technology.

[0004] To achieve the above objectives, this application discloses the following technical solutions:

[0005] A conference summary generation system based on natural language processing, comprising:

[0006] A speaker identification module, which is used to obtain speaker information, including the speaker's identity and its corresponding rank;

[0007] A text processing module, the text processing module is used to obtain the meeting record and generate a record text based on the speaker identity in the speaker information, the record text including the speaker identity and the speech content corresponding to the speaker;

[0008] A first keyword generation module, the first keyword generation module is used to generate a first keyword set according to the conference theme based on natural language processing technology;

[0009] A first summary generation module, the first summary generation module is used to generate a first summary corresponding to the record text according to the first keyword set based on natural language processing technology;

[0010] A second keyword generation module, the second keyword generation module is used to generate a second keyword set based on the first keyword set and assign a value to the job level corresponding to the spokesperson as a reference;

[0011] A second summary generating module, the second summary generating module is used to generate a second summary corresponding to the record text according to the second keyword set based on natural language processing technology;

[0012] Either the first summary or the second summary is selected as the meeting summary of the current meeting.

[0013] Preferably, the speaker identity recognition module is further used to number the speaker identities and their corresponding job levels in the speaker information, and the numbering results are 1, 2, 3...n.

[0014] Preferably, the text processing module is also used to identify the type of the meeting minutes, which includes any one or more of video, audio or text. When the type of the meeting minutes is video or audio, speech recognition is performed on the meeting minutes to generate the record text; when the type of the meeting minutes is text, the record text is directly generated.

[0015] Preferably, the first keyword generation module is used to generate a first keyword set according to the conference theme based on natural language processing technology, including:

[0016] The conference theme is obtained. The first keyword generation module generates a first candidate keyword set related to the conference theme based on natural language processing technology, numbers the keywords, and the numbering results are 1, 2, 3, ... i, and obtains the correlation degree F between the keywords and the conference theme. i ;

[0017] In the recorded text, the number of occurrences of the keywords in the first candidate keyword set is counted, and the counting result is defined as G i ;

[0018] Calculate the first keyword score J i , J i =F i *G i , score the first keyword J i An analysis is performed to extract keywords that meet the conditions from the first candidate keyword set to generate the first keyword set.

[0019] Preferably, the degree of association between the acquired keyword and the conference theme F i The first keyword generation module obtains the degree of association between the keyword and the conference theme based on a preset degree of association table. i .

[0020] As a preferred embodiment, the scoring of the first keyword J i The analysis includes:

[0021] Preset the first keyword evaluation threshold M, when J i When ≥M, it means that the keyword numbered i can be used as the keyword of the current meeting, it is defined as the first candidate keyword, and the first candidate keyword set is generated; otherwise, it means that the keyword numbered i cannot be used as the keyword of the current meeting;

[0022] The first keyword generating module generates the first candidate keyword set according to the first keyword score J i Sort from high to low, and take the first keyword score J in the first candidate keyword i The highest A keywords are used as keywords for the current meeting, and the first keyword set is generated.

[0023] Preferably, the second keyword generation module is used to generate the second keyword set based on the first keyword set by assigning a value to the job level corresponding to the spokesperson as a reference, and the second keyword set includes:

[0024] The second keyword generation module splits the recorded text according to the speaker and the corresponding speech content to obtain the speaker text;

[0025] In the speaker text, the number of occurrences of the keywords in the first keyword set is counted, and the counting result is defined as E i , and the counting result E i Conduct analysis;

[0026] The second keyword generation module obtains the corresponding speaker weight D based on the corresponding job level in the speaker information. n ;

[0027] Calculate the second keyword score K i , K i =J i *E i *D n , score the second keyword K i An analysis is performed to extract keywords that meet the conditions from the second candidate keyword set to generate the second keyword set.

[0028] Preferably, the counting result E i The analysis includes:

[0029] Preset the second keyword evaluation threshold R, when E i >R, it means that the keyword numbered i can be used as a keyword in the speech of the speaker numbered n, otherwise it means that the keyword numbered i cannot be used as a keyword in the speech of the speaker numbered n.

[0030] As a preference, the scoring of the second keyword K i The analysis includes:

[0031] Preset the third keyword evaluation threshold N, when K iWhen ≥N, it means that the keyword numbered i can be used as the keyword of the current meeting, it is defined as the second candidate keyword, and the second candidate keyword set is generated; otherwise, it means that the keyword numbered i cannot be used as the keyword of the current meeting;

[0032] The second keyword generation module generates the second candidate keyword set according to the second keyword score K i Sort from high to low, and take the second keyword score K in the second candidate keyword i The highest B keywords are used as keywords for the current meeting, and the second keyword set is generated.

[0033] Beneficial effects: The conference summary generation system based on natural language processing of the present application realizes the conversion of conference records in video or audio format through the speaker identity recognition module and the text processing module, thereby simplifying the process of organizing conference records; realizes the function of extracting conference summaries according to conference themes through the first keyword generation module and the first summary generation module; realizes keyword optimization based on the speaker's rank through the second keyword generation module and the second summary generation module, thereby improving the quality of conference summaries and avoiding unnecessary waste of manpower and material resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 This is a structural block diagram of a conference summary generation system based on natural language processing provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] The technical solutions in the embodiments of the present application are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0037] In this article, the term "comprising" is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of more restrictions, the elements defined by the sentence "comprising..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0038] This embodiment discloses Figure 1 A conference summary generation system based on natural language processing is shown, comprising:

[0039] The speaker identity recognition module is used to obtain the speaker information, which includes the speaker identity and its corresponding rank.

[0040] The text processing module is used to obtain the meeting minutes and generate the record text based on the speaker identity in the speaker information. The record text includes the speaker identity and the corresponding speech content of the speaker.

[0041] The first keyword generation module is used to generate a first keyword set according to the conference theme based on natural language processing technology.

[0042] The first summary generation module is used to generate a first summary corresponding to the record text according to the first keyword set based on natural language processing technology.

[0043] The second keyword generation module is used to generate a second keyword set based on the first keyword set, assigning the corresponding rank of the speaker as a reference; wherein, assigning the corresponding rank of the speaker as a reference refers to a preset rank assignment table, assigning the rank of the speaker from high to low. In a simple example, for example, the speakers are divided into the first, second and third levels from high to low, and the following rank assignment table can be used:

[0044]

[0045] The second summary generation module is used to generate a second summary corresponding to the record text according to the second keyword set based on natural language processing technology.

[0046] Either the first summary or the second summary is selected as the meeting summary of the current meeting.

[0047] Specifically, the speaker identity recognition module is further used to number the speaker identities and their corresponding job levels in the speaker information, and the numbering results are 1, 2, 3...n.

[0048] Specifically, the text processing module is also used to identify the type of meeting minutes, which may be any one or more of video, audio or text. When the type of the meeting minutes is video or audio, speech recognition is performed on the meeting minutes to generate the record text; when the type of the meeting minutes is text, the record text is directly generated.

[0049] Specifically, the first keyword generation module is used to generate the first keyword set according to the conference theme based on the natural language processing technology, including:

[0050] The conference theme is obtained. The first keyword generation module generates a first candidate keyword set related to the conference theme based on natural language processing technology, numbers the keywords, and the numbering results are 1, 2, 3, ... i, and obtains the correlation degree F between the keywords and the conference theme. i .

[0051] In the recorded text, the number of occurrences of the keywords in the first candidate keyword set is counted, and the counting result is defined as G i .

[0052] Calculate the first keyword score J i , J i =F i *G i , score the first keyword J i An analysis is performed to extract keywords that meet the conditions from the first candidate keyword set to generate a first keyword set.

[0053] Specifically, obtain the correlation degree F between the keyword and the conference topic i The first keyword generation module obtains the correlation degree F between the keyword and the conference theme based on the preset correlation degree table. i .

[0054] Specifically, score the first keyword J i The analysis includes:

[0055] Preset the first keyword evaluation threshold M, when J i ≥M, it means that the keyword numbered i can be used as the keyword of the current meeting, it is defined as the first candidate keyword, and the first candidate keyword set is generated. Otherwise, it means that the keyword numbered i cannot be used as the keyword of the current meeting.

[0056] The first keyword generation module generates the first candidate keyword set according to the first keyword score J i Sort from high to low, and take the first keyword score J in the first candidate keyword i The highest A keywords are used as keywords for the current meeting, and a first keyword set is generated.

[0057] Based on the above, this embodiment verifies the accuracy of the extracted keywords, ensuring the accuracy when the conference summary is generated based on the keywords.

[0058] Specifically, the second keyword generation module is used to assign a value to the job level corresponding to the spokesperson as a reference based on the first keyword set, and generate the second keyword set including:

[0059] The second keyword generation module splits the recorded text according to the speaker and the corresponding speech content to obtain the speaker text.

[0060] In the speaker text, count the number of occurrences of the keywords in the first keyword set, and define the count result as E i , and the counting result E i Conduct analysis.

[0061] The second keyword generation module obtains the corresponding speaker weight D based on the corresponding job level in the speaker information. n .

[0062] Calculate the second keyword score K i , K i =J i *E i *D n , score the second keyword K i An analysis is performed to extract keywords that meet the conditions from the second candidate keyword set to generate a second keyword set.

[0063] In this regard, this embodiment performs keyword optimization under specific circumstances to improve the quality of the conference abstract.

[0064] Specifically, for the counting result E i The analysis includes:

[0065] Preset the second keyword evaluation threshold R, when E i >R, it means that the keyword numbered i can be used as a keyword in the speech of the speaker numbered n, otherwise it means that the keyword numbered i cannot be used as a keyword in the speech of the speaker numbered n.

[0066] In a simple example, for example, the second keyword evaluation threshold R=1, the keyword is "construction", and the speech content of the speaker with number n=1 and a first-level rank is "keep up the momentum". i =0, does not satisfy E i > R, the keyword "construction" cannot be used as a keyword in the speech of the speaker numbered 1; the speech content of the speaker numbered n=2 and the second-level position is "adopting this construction method improves construction efficiency" at this time E i=2, satisfying E i >R, the keyword "construction" can be used as a keyword in the speech of speaker number 2.

[0067] Specifically, the second keyword is scored K i The analysis includes:

[0068] Preset the third keyword evaluation threshold N, when K i ≥N, it means that the keyword numbered i can be used as the keyword of the current meeting, it is defined as the second candidate keyword, and a second candidate keyword set is generated. Otherwise, it means that the keyword numbered i cannot be used as the keyword of the current meeting.

[0069] The second keyword generation module generates the second candidate keyword set according to the second keyword score K i Sort from high to low and take the second keyword score K in the second candidate keyword i The highest B keywords are used as keywords for the current meeting, and a second keyword set is generated.

[0070] In summary, the conference summary generation system based on natural language processing in this embodiment realizes the conversion of conference records in video or audio format through the speaker identity recognition module and the text processing module, thereby simplifying the process of organizing conference records; realizes the function of extracting conference summaries according to conference topics through the first keyword generation module and the first summary generation module; realizes keyword optimization based on the speaker's rank through the second keyword generation module and the second summary generation module, thereby improving the quality of conference summaries and avoiding unnecessary waste of manpower and material resources.

[0071] Finally, it should be noted that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A conference summary generation system based on natural language processing, characterized in that: include: A speaker identification module, which is used to obtain speaker information, including the speaker's identity and its corresponding rank; A text processing module, the text processing module is used to obtain the meeting record and generate a record text based on the speaker identity in the speaker information, the record text including the speaker identity and the speech content corresponding to the speaker; A first keyword generation module, the first keyword generation module is used to generate a first keyword set according to the conference theme based on natural language processing technology; A first summary generation module, the first summary generation module is used to generate a first summary corresponding to the record text according to the first keyword set based on natural language processing technology; A second keyword generation module, the second keyword generation module is used to generate a second keyword set based on the first keyword set and assign a value to the job level corresponding to the spokesperson as a reference; A second summary generating module, the second summary generating module is used to generate a second summary corresponding to the record text according to the second keyword set based on natural language processing technology; Select either the first abstract or the second abstract as the meeting abstract of the current meeting; The speaker identity recognition module is also used to number the speaker identity and the corresponding rank in the speaker information, and the numbering result is: ; The text processing module is further used to identify the type of the meeting record, which includes any one or more of video, audio or text. When the type of the meeting record is video or audio, speech recognition is performed on the meeting record to generate the record text; when the type of the meeting record is text, the record text is directly generated; The first keyword generation module is used to generate a first keyword set according to the conference theme based on natural language processing technology, including: The conference theme is obtained. The first keyword generation module generates a first candidate keyword set related to the conference theme based on natural language processing technology and numbers the keywords. The numbering result is , and obtain the degree of relevance between the keywords and the conference theme F i ; In the recorded text, the number of occurrences of the keywords in the first candidate keyword set is counted, and the count result is defined as G i ; Calculate the first keyword score J i , , score the first keyword J i Performing analysis, extracting keywords that meet the conditions from the first candidate keyword set, and generating the first keyword set; The second keyword generation module is used to generate the second keyword set based on the first keyword set by assigning a value to the job level corresponding to the spokesperson as a reference, and includes: The second keyword generation module splits the recorded text according to the speaker and the corresponding speech content to obtain the speaker text; In the speaker text, the number of occurrences of the keywords in the first keyword set is counted, and the count result is defined as , and the counting results Conduct analysis; The second keyword generation module obtains the corresponding speaker weight based on the corresponding job level in the speaker information ; Calculate the second keyword score , , score the second keyword An analysis is performed to extract keywords that meet the conditions from the second candidate keyword set to generate the second keyword set.

2. The conference summary generation system based on natural language processing according to claim 1 is characterized in that: The degree of relevance between the acquired keywords and the conference theme F i The first keyword generation module obtains the degree of association between the keyword and the conference theme based on a preset association degree table. F i .

3. The conference summary generation system based on natural language processing according to claim 1 is characterized in that: The scoring of the first keyword J i The analysis includes: Preset the first keyword evaluation threshold M ,when J i ≥M , indicating that the number is i The keyword can be used as the keyword of the current meeting, defined as the first candidate keyword, and the first candidate keyword set is generated. Otherwise, it means that the number is i The keywords cannot be used as keywords for the current meeting; The first keyword generating module generates the first candidate keyword set according to the first keyword score. J i Sort from high to low and take the first keyword score in the first candidate keyword set J i high A keywords as keywords for the current meeting, and generate the first keyword set.

4. The conference summary generation system based on natural language processing according to claim 1 is characterized in that: The pair counting results E i The analysis includes: Preset the second keyword evaluation threshold R ,when E i ＞R , indicating that the number is i The keywords are in the number n can be used as a keyword in the speech of the speaker, otherwise it means that the number is i The keywords are in the number n It cannot be used as a keyword in the speaker's speech.

5. The conference summary generation system based on natural language processing according to claim 1 is characterized in that: The scoring of the second keyword K i The analysis includes: Preset the third keyword evaluation threshold N ,when K i ≥N , indicating that the number is i The keyword can be used as the keyword of the current meeting, defined as the second candidate keyword, and generate the second candidate keyword set, otherwise it means numbered i The keywords cannot be used as keywords for the current meeting; The second keyword generating module generates the second candidate keyword set according to the second keyword score. K i Sort from high to low, and take the second keyword score of the second candidate keyword K i high B keywords as keywords for the current meeting, and generate the second keyword set.

Images