CN114861657A

CN114861657A - Conference key sentence extraction method and device

Info

Publication number: CN114861657A
Application number: CN202210542540.7A
Authority: CN
Inventors: 郭馨泽; 李长亮
Original assignee: Beijing Kingsoft Digital Entertainment Co Ltd
Current assignee: Beijing Kingsoft Digital Entertainment Co Ltd
Priority date: 2022-05-18
Filing date: 2022-05-18
Publication date: 2022-08-05

Abstract

The application provides a meeting key sentence extracting method and device, wherein the meeting key sentence extracting method comprises the following steps: the method comprises the steps of obtaining a conference text to be processed and a target extraction type of the conference text to be processed, identifying the conference text to be processed according to the target extraction type, and obtaining a conference key sentence corresponding to the target extraction type. By identifying the conference text to be processed according to the target extraction type, the accuracy of the acquired conference key sentence is improved.

Description

Conference key sentence extraction method and device

Technical Field

The application relates to the technical field of artificial intelligence, in particular to a meeting key sentence extraction method. The application also relates to a meeting key sentence extracting device, a computing device and a computer readable storage medium.

Background

With the development of internet technology, text reasoning is more and more dependent on the internet, the text reasoning is a process of analyzing, understanding, extracting and the like of a text, and the text reasoning can help people to perform operations such as text understanding, aggregation analysis, abstract generation, information extraction and the like. Therefore, the text reasoning has been widely applied to various fields of daily life of people.

In the prior art, a text is generally divided into a plurality of groups of words or sentences, a graph model is established based on the words or sentences obtained by division, and important components in the text are sequenced by using a voting mechanism, so that the extraction of key contents of the text is realized. However, the text key content obtained by the above method has poor accuracy due to the excessive amount of text.

Disclosure of Invention

In view of this, the embodiment of the present application provides a method for extracting a meeting key sentence, so as to solve technical defects in the prior art. The embodiment of the application also provides a meeting key sentence extracting device, a computing device and a computer readable storage medium.

According to a first aspect of the embodiments of the present application, a method for extracting a meeting key sentence is provided, which includes:

acquiring a to-be-processed conference text and a target extraction type of the to-be-processed conference text;

and identifying the conference text to be processed according to the target extraction type to obtain the conference key sentence corresponding to the target extraction type.

According to a second aspect of the embodiments of the present application, there is provided a meeting key sentence extraction apparatus, including:

the acquisition module is configured to acquire the conference text to be processed and the target extraction type of the conference text to be processed;

and the processing module is configured to identify the conference text to be processed according to the target extraction type, and obtain the conference key sentence corresponding to the target extraction type.

According to a third aspect of embodiments herein, there is provided a computing device comprising:

a memory and a processor;

the memory is used for storing computer executable instructions, and the processor realizes the steps of the conference key sentence extraction method when executing the computer executable instructions.

According to a fourth aspect of embodiments of the present application, there is provided a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the meeting key sentence extraction method.

According to a fifth aspect of the embodiments of the present application, there is provided a chip storing a computer program, which when executed by the chip, implements the steps of the meeting key sentence extracting method.

According to the conference key sentence extraction method, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is identified according to the target extraction type, and the conference key sentence corresponding to the target extraction type is obtained. By identifying the conference text to be processed according to the target extraction type, the accuracy of the acquired conference key sentence is improved.

Drawings

Fig. 1 is a schematic structural diagram of a meeting key sentence extraction system according to an embodiment of the present application;

fig. 2 is a flowchart of a first method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 3 is a flowchart of a second method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 4 is a flowchart of a third method for extracting a key sentence of a meeting according to an embodiment of the present application;

fig. 5 is a flowchart of a fourth method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 6 is a flowchart of a fifth method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 7 is a flowchart of a sixth method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 8 is a flowchart illustrating training of a sentence classification model in a method for extracting a key sentence of a conference according to an embodiment of the present application;

fig. 9 is a flowchart illustrating a sentence classification model in another method for extracting a key sentence in a meeting according to an embodiment of the present application;

fig. 10 is a flowchart of a seventh method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 11 is a flowchart of an eighth method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 12 is a flowchart of a ninth method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 13 is a schematic diagram of a method for extracting a meeting key sentence according to an embodiment of the present application;

fig. 14 is a schematic structural diagram of a conference key sentence extraction apparatus according to an embodiment of the present application;

fig. 15 is a block diagram of a computing device according to an embodiment of the present application.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. This application is capable of implementation in many different ways than those herein set forth and of similar import by those skilled in the art without departing from the spirit and scope of this application, and thus this application is not limited to the specific implementations disclosed below.

The terminology used in the one or more embodiments of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of the present application. As used in one or more embodiments of the present application and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used in one or more embodiments of the present application refers to and encompasses any and all possible combinations of one or more of the associated listed items.

It will be understood that, although the terms first, second, etc. may be used herein in one or more embodiments of the present application to describe various information, these information should not be limited by these terms. These terms are only used to distinguish one type of information from another. For example, a first aspect may be termed a second aspect, and, similarly, a second aspect may be termed a first aspect, without departing from the scope of one or more embodiments of the present application.

First, the noun terms to which one or more embodiments of the present specification relate are explained.

Artificial Intelligence (AI) refers to the ability of an engineered (i.e., designed and manufactured) system to perceive the environment, as well as the ability to acquire, process, apply, and represent knowledge. Natural language processing, robotics, and computer vision are the most popular three industries for artificial intelligence. The development conditions of key technologies in the field of artificial intelligence comprise key technologies such as machine learning, knowledge maps, natural language processing, computer vision, human-computer interaction, biological feature recognition, virtual reality/augmented reality and the like. Natural language processing is an important direction in the fields of computer science and artificial intelligence, and various theories and methods for realizing effective communication between people and computers by using natural language are researched, and the fields related to the natural language processing mainly comprise machine translation, machine reading understanding, text extraction and the like.

TextRank: the TextRank algorithm is a graph-based ranking algorithm for text. The basic idea is derived from a PageRank algorithm, a text is divided into a plurality of composition units (words and sentences), a graph model is established, important components in the text are sequenced by using a voting mechanism, and keyword extraction and abstract can be realized only by using the information of a single document. The TextRank does not need to perform learning training on a plurality of documents in advance, and is widely applied due to simplicity and effectiveness.

BERT (bidirectional Encoder retrieval from transformations): the language model is a pre-training language model, and is trained by constructing a word prediction task and a next sentence prediction task to learn language knowledge.

BERTSUM (Fine-tune BERT for extraction Summarization): and using the sentence representation output by the BERT, and capturing the document characteristics through a network layer to classify the sentences so as to judge whether the sentences are the sentences in the abstract or not.

In the application, a meeting key sentence extracting method is provided. The present application also relates to a meeting key sentence extraction apparatus, a computing device, and a computer-readable storage medium, which are described in detail in the following embodiments one by one.

In practical application, an unsupervised method can be adopted for extracting key sentences, for example, a TextRank and similarity calculation method is adopted, a text is divided into a plurality of groups of words or sentences, a graph model is established based on the words or sentences obtained by division, and important components in the text are sequenced by using a voting mechanism, so that the extraction of the key contents of the text is realized. A classification model may also be used to classify key sentences, such as BERTSUM, however, due to the large amount of text, the accuracy of the key content of the text obtained by the above method is poor.

In order to improve the accuracy of text key content extraction, the application provides a conference key sentence extraction method, which includes the steps of obtaining a to-be-processed conference text and a target extraction type of the to-be-processed conference text, identifying the to-be-processed conference text according to the target extraction type, and obtaining a conference key sentence corresponding to the target extraction type. By identifying the conference text to be processed according to the target extraction type, the accuracy of the acquired conference key sentence is improved.

Referring to fig. 1, fig. 1 shows a schematic structural diagram of a meeting key sentence extraction system provided in an embodiment of the present application.

The method comprises two execution bodies, namely a server 102 and a terminal 104, wherein the server 102 is used for extracting the meeting key sentence, and the terminal 104 is used for providing the meeting text to be processed and the target extraction type for the server 102.

That is to say, the server 102 obtains the to-be-processed conference text and the target extraction type of the to-be-processed conference text from the terminal 104, identifies the to-be-processed conference text according to the target extraction type, obtains the conference key sentence corresponding to the target extraction type, and feeds back the conference key sentence corresponding to the target extraction type to the terminal 104.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is identified according to the target extraction type, and the conference key sentence corresponding to the target extraction type is obtained. By identifying the conference text to be processed according to the target extraction type, the accuracy of the acquired conference key sentence is improved.

Referring to fig. 2, fig. 2 is a flowchart illustrating a first method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S202: and acquiring the meeting text to be processed and the target extraction type of the meeting text to be processed.

Specifically, the meeting refers to an assembly of multiple meetings and a meeting and follows a certain agenda, the meeting text refers to a text in which relevant matters of the meeting are recorded by characters, the meeting text to be processed refers to a meeting text which needs key sentence extraction, the target extraction type refers to a type in which a user wants to extract a meeting key sentence in the meeting text to be processed, the target extraction type may be one or multiple, and the target extraction type is selected specifically according to actual conditions.

In one or more embodiments of the present application, because the conference text includes a plurality of different text types, in the process of extracting the conference key sentence, the target extraction type of the conference text to be processed may be obtained, and the conference key sentence is extracted by using the target extraction type of the conference text to be processed.

It should be noted that the target extraction type includes at least one of a topic type, a conclusion type, a to-be-done type and a type containing key information, where the topic refers to the central content discussed in the meeting, and the topic of the meeting is generally a specific meeting for a certain event or a certain activity, and discussion research is performed around the event. The conclusion is that the ending part of the conference is the closing word made around the conference. To do refers to the work to be done after the meeting. The key information refers to key content emphasized in the conference, and is specifically selected according to actual situations, which is not limited in this embodiment of the present application.

Illustratively, the obtained meeting text to be processed is that the meeting notification conveys the spirit of the education work and the key point content of the education work, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. And acquiring the target extraction type of the conference text to be processed as a to-be-done type.

In practical application, there are various ways of obtaining the to-be-processed conference text and the target extraction type of the to-be-processed conference text, and the method is specifically selected according to an actual situation, and the embodiment of the present application does not limit this.

In a possible implementation manner, the to-be-processed conference text and the target extraction type of the to-be-processed conference text can be directly obtained.

In another possible implementation manner, conference audio or video sent by the terminal may be received, and the conference audio or the conference video is converted into a conference text by using a preset conversion tool, where the preset conversion tool is a tool that can convert audio or video into text, and is specifically selected according to an actual situation, and this is not limited in this embodiment of the present application.

Step S204: and identifying the conference text to be processed according to the target extraction type to obtain the conference key sentence corresponding to the target extraction type.

Specifically, the meeting key sentence refers to a meeting sentence corresponding to the target extraction type in the meeting text.

In practical application, according to the target extraction type, there are various ways of identifying the conference text to be processed, which are specifically selected according to practical situations, and this is not limited in this embodiment of the present application.

In a possible implementation manner, a pre-set conference key sentence template corresponding to the target extraction type can be used to identify the conference text to be processed, so as to obtain the conference key sentence corresponding to the target extraction type.

In another possible implementation manner, a pre-trained model can be used to identify the conference text to be processed, so as to obtain the conference key sentence corresponding to the target extraction type.

Illustratively, the above example is continued, the obtained meeting text to be processed is that "the meeting notification conveys the spirit of the education work and the key point content of the education work, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. The method comprises the steps of acquiring a target extraction type of a meeting text to be processed as a to-be-made type, acquiring a meeting key sentence template corresponding to the target extraction type of the to-be-made type as 'work below arrangement of people', identifying the meeting text to be processed by using the meeting key sentence template, acquiring a meeting key sentence corresponding to the meeting text to be processed as 'work below arrangement of the meeting', fulfilling diligence and frugal tasks of students, moving a teaching worker restaurant to a second floor, and completely supplying the meeting text to students in the first floor, so as to solve the problem of crowded dining of the students. ".

The core of the embodiment of the application lies in identifying the conference text to be processed, the process of identifying the conference text to be processed is basically the same for different target extraction types, and the process of identifying the conference text to be processed is described in detail below.

Referring to fig. 3, fig. 3 is a flowchart illustrating a second method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S302: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

The expression of step S302 is the same as that of step S202, and the description thereof is omitted here.

Step S304: and obtaining a target key sentence template corresponding to the target extraction type according to the target extraction type.

Specifically, the target key sentence template refers to a key sentence template corresponding to a target extraction type of the conference text to be processed, and there are various target key sentence templates, which are selected according to actual situations.

In one or more embodiments of the application, after the meeting text to be processed and the target extraction type of the meeting text to be processed are obtained, a target key sentence template corresponding to the target extraction type can be obtained, and a meeting key sentence corresponding to the target extraction type in the meeting text to be processed is extracted by using the target key sentence template.

In a possible implementation manner, a key sentence template corresponding to the target extraction type may be directly searched in a preset key sentence template library, and the key sentence template corresponding to the target extraction type is used as the target key sentence template.

Illustratively, the obtained meeting text to be processed is that the meeting notification conveys the spirit of the education work and the key point content of the education work, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. "the target extraction type of the obtained conference text to be processed is" subject type ", and a target key sentence template corresponding to the target extraction type" subject type "is searched in a preset key sentence template library to be" the conference notification is conveyed ".

In another possible implementation manner, a preset dependency syntax template may be obtained, and the content in the dependency syntax template is replaced with the target extraction type to obtain the target key sentence template.

In particular, dependency syntax (DP) can reveal its syntax structure by analyzing dependencies between components within a language unit. Dependency syntax can also be used to parse and identify grammatical components "predicate object", "predicate complement" in a sentence, and to parse the relationships between the components.

Illustratively, the preset dependency syntax template is acquired as ' noun + be verb + clause ', the target extraction type of the conference text to be processed is ' conclusion type ', and the target extraction type is used for performing keyword replacement on the dependency syntax template to obtain a target key sentence template as ' conclusion is.

Step S306: and matching the meeting text to be processed with the target key sentence template to obtain a meeting key sentence corresponding to the target extraction type.

Specifically, there are various ways to match the conference text to be processed with the target key sentence template, which are specifically selected according to actual situations, and this is not limited in this embodiment of the present application.

In a possible implementation manner, the content of the target key template can be directly searched in the conference text to be processed, and the conference key sentence corresponding to the target extraction type is obtained.

In another possible implementation manner, the text to be processed may be segmented, the similarity between each segmented word and the target key sentence template is calculated, and the conference key sentences corresponding to the target extraction types are obtained by sorting according to the similarity.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the target key sentence template corresponding to the target extraction type is obtained according to the target extraction type, the conference text to be processed is matched with the target key sentence template, the conference key sentence corresponding to the target extraction type is obtained, and the accuracy of the obtained conference key sentence is improved.

Referring to fig. 4, fig. 4 shows a flowchart of a third method for extracting a key sentence of a meeting, which may specifically include the following steps:

step S402: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

The expression of step S402 is the same as that of step S202, and the description thereof is omitted here.

Step S404: and acquiring a preset key sentence template base, wherein the key sentence template base comprises a plurality of types of key sentence templates.

Step S406: and searching a first target key sentence template corresponding to the target extraction type in a preset key sentence template base.

Step S408: and matching the meeting text to be processed with the first target key sentence template to obtain a meeting key sentence corresponding to the target extraction type.

Specifically, the preset key sentence template library includes multiple types of key sentence templates, including but not limited to a topic type key sentence template, a to-be-made type key sentence template, a conclusion type key sentence template, and a key sentence template including a key information type, which are specifically selected according to actual situations, and this is not limited in any way in the embodiments of the present application.

In one or more embodiments of the present application, after obtaining a to-be-processed conference text and a target extraction type of the to-be-processed conference text, a target key sentence template corresponding to the target extraction type may be obtained in a preset key sentence template base, and a conference key sentence corresponding to the target extraction type in the to-be-processed conference text is obtained by using the target key sentence template.

Illustratively, the obtained preset key sentence template library comprises a theme type key sentence template, wherein the meeting notice conveys that the meeting is completed, a to-be-completed type key sentence template and a conclusion type key sentence template, wherein the meeting conclusion is that the meeting is completed. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. "the target extraction type of the conference text to be processed is" topic type ", a first target key sentence template corresponding to the target extraction type" topic type "is searched in a preset key sentence template library to be" the conference notification conveys. ", the conference text to be processed is matched with the first target key sentence template, and the obtained conference key sentence corresponding to the target extraction type is" the conference notification conveys the spirit of the education work and the key point content of the education work ".

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the preset key sentence template base is obtained, wherein the key sentence template base comprises multiple types of key sentence templates, the first target key sentence template corresponding to the target extraction type is searched in the preset key sentence template base, the conference text to be processed and the first target key sentence template are matched, the conference key sentence corresponding to the target extraction type is obtained, and the accuracy of the obtained conference key sentence is improved.

Referring to fig. 5, fig. 5 is a flowchart illustrating a fourth method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S502: and acquiring the meeting text to be processed and the target extraction type of the meeting text to be processed.

The expression of step S502 is the same as that of step S202, and the description thereof is omitted here.

Step S504: and acquiring a preset dependency syntax template.

Step S506: and performing keyword replacement on the dependency syntax template based on the target extraction type to obtain a second target key sentence template.

Step S508: and matching the meeting text to be processed with the second target key sentence template to obtain a meeting key sentence corresponding to the target extraction type.

Illustratively, since the dependency syntax can analyze grammatical components in a sentence, the dependency syntax can be utilized to analyze a sentence "three-dimensional research on a high bridge and active exploration new mechanism supporting city a" to obtain a core predicate of the sentence "proposed", a subject of the sentence "three-dimensional research", and an object of the "proposed" is "supporting city a. When "investigate," is the time-like phrase "proposed," the object of "support" is "explore new mechanisms.

In one or more embodiments of the present application, after obtaining a to-be-processed conference text and a target extraction type of the to-be-processed conference text, a preset dependency syntax template may be obtained, and a conference key sentence corresponding to the target extraction type in the to-be-processed conference text is obtained by using the dependency syntax template.

In practical applications, two dependency syntax templates, namely "noun + be verb + clause" and "verb + noun or clause", may be preset, where the noun and the verb may be replaced according to different keywords of the target extraction type. For example: topic type sentence extraction uses a dependency syntax match of "topic" + "is" + clauses. In addition, sentence template matching of "future time + verb" is also taken for the type of sentence to be made.

Illustratively, the preset dependency syntax template is acquired as ' noun + be verb + clause ', the conference text to be processed is ' the notification of the conference conveys the spirit of the education work and the key point content of the education work, the conclusion of the conference is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large scale within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. "the target extraction type of the conference text to be processed is" conclusion type ", the noun in the dependency syntax template is replaced with" conclusion ", the verb be is replaced with" yes ", the second target key sentence template is" conclusion yes ", the conference text to be processed is matched with the second target key sentence template, and the conference key sentence corresponding to the target extraction type is obtained.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the preset dependency syntax template is obtained, the dependency syntax template is subjected to keyword replacement based on the target extraction type, the second target key sentence template is obtained, the conference text to be processed and the second target key sentence template are matched, the conference key sentence corresponding to the target extraction type is obtained, and the accuracy of the obtained conference key sentence is improved.

Referring to fig. 6, fig. 6 is a flowchart illustrating a fifth method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S602: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

The expression of step S602 is the same as that of step S202, and the description of this embodiment is omitted.

Step S604: and inputting the conference text to be processed into a pre-trained sentence classification model to obtain the sentence type of each conference sentence in the conference text to be processed.

Step S606: and acquiring the conference key sentence corresponding to the target extraction type based on the sentence type of each conference sentence.

Specifically, the pre-trained sentence classification model refers to a pre-trained model that can classify sentences.

In one or more embodiments of the application, after the to-be-processed conference text and the target extraction type of the to-be-processed conference text are obtained, the to-be-processed conference text can be input into a pre-trained sentence classification model, and a conference key sentence corresponding to the target extraction type in the to-be-processed conference text is obtained.

It should be noted that the to-be-processed conference text is input into a pre-trained sentence classification model, and sentence types of each conference sentence in the to-be-processed conference text are obtained, where each sentence type may include one conference sentence or may include a plurality of conference sentences.

In a possible implementation manner, a to-be-processed conference text is input into a pre-trained sentence classification model, sentence types of conference sentences in the to-be-processed conference text are obtained, each sentence type corresponds to one conference sentence, and at this time, a conference sentence corresponding to a target extraction type is directly used as a conference key sentence corresponding to the target extraction type of the to-be-processed conference text.

Illustratively, the obtained meeting text to be processed is that the meeting notification conveys the spirit of the education work and the key point content of the education work, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. The meeting sentence type of meeting sentences is a conclusion type, the meeting sentences is a sentence type, the meeting sentences are arranged for the following work, the orderly and economic tasks of students are implemented, the teacher dining room is moved to the second floor, all the first floor dining rooms are used by the students, and the sentence type for solving the problem that the students have a crowded meal is a 'to-be-done type'. The target extraction type of the conference text to be processed is obtained as a conclusion type, and the conclusion of the conference sentence which is the same as the target extraction type and is called and reported to the leader by each unit in time, and the leader is required to work in a large amount within the scope of the authority is used as the conference key sentence corresponding to the target extraction type of the conference text to be processed.

In another possible implementation manner, the conference text to be processed is input into a pre-trained sentence classification model, and sentence types of each conference sentence in the conference text to be processed are obtained, each sentence type corresponds to a plurality of conference sentences, at this time, a conference sentence meeting a preset condition among the plurality of conference sentences corresponding to the target extraction type can be used as a conference key sentence corresponding to the target extraction type of the conference text to be processed.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is input into the pre-trained sentence classification model, the sentence type of each conference sentence in the conference text to be processed is obtained, the conference key sentence corresponding to the target extraction type is obtained based on the sentence type of each conference sentence, and the accuracy of the obtained conference key sentence is improved.

Referring to fig. 7, fig. 7 is a flowchart illustrating a sixth method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S702: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

Step S704: and inputting the conference text to be processed into a pre-trained sentence classification model to obtain the sentence type of each conference sentence in the conference text to be processed.

The expressions of step S702 and step S704 are the same as those of step S602 and step S604, and the description thereof is omitted here.

Step S706: and classifying the conference sentences according to the sentence types.

Step S708: and acquiring a plurality of conference sentences corresponding to the target extraction type, and taking the conference sentences meeting preset conditions as conference key sentences corresponding to the target extraction type.

Specifically, the preset condition is a preset condition that can determine a conference key sentence from a plurality of conference sentences, and is specifically selected according to an actual situation, which is not limited in this embodiment of the present application.

Illustratively, the meeting text to be processed is acquired as that the meeting notification conveys the spirit of the education work and the key point content of the education work, the subject is to care students and promote the education development, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. The method comprises the steps of inputting a text of a meeting to be processed into a pre-trained sentence classification model, obtaining a meeting statement, wherein the sentence type of a meeting notice that the spirit of education work and the key point content of the education work are transmitted is a topic type, the topic of the meeting statement is a subject for caring students and promoting education development is a topic type, the conclusion of the meeting statement is that each unit needs to ask for and report work to leaders in time, the sentence type of the leading needs to work in a courage range is a conclusion type, the meeting statement is arranged in the following work, the diligence and economic tasks of students are implemented, a teacher restaurant is moved to the second floor, all first floor restaurants are used by the students, and the sentence type of the problem that the students have a crowded meal is a to be made is a to be solved. The target extraction type for obtaining the conference text to be processed is a theme type, and correspondingly, two conference sentences corresponding to the target extraction type are obtained, namely the meeting notification conveys the spirit of the education work and the key content of the education work, and the theme is concerned with students and promotes education development. At this time, the preset condition is obtained that the number of the meeting key sentence words is more than 15, and the meeting sentence meeting the preset condition, namely the meeting notice conveys the spirit of the education work and the key content of the education work, is used as the meeting key sentence corresponding to the target extraction type of the meeting text to be processed.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is input into a pre-trained sentence classification model, the sentence type of each conference sentence in the conference text to be processed is obtained, each conference sentence is classified according to the sentence type, a plurality of conference sentences corresponding to the target extraction type are obtained, the conference sentences meeting preset conditions are used as conference key sentences corresponding to the target extraction type, and the accuracy of the obtained conference key sentences is improved.

The core of the above embodiment of the present application lies in the training process of the sentence classification model, which is described in detail below.

Referring to fig. 8, fig. 8 shows a training flowchart of a sentence classification model in a method for extracting a key sentence of a conference, which is provided by an embodiment of the present application, and specifically includes the following steps:

step S802: obtaining a sample set, wherein the sample set comprises a plurality of types of conference samples, and each conference sample carries a type label.

In one or more embodiments of the present application, when a sentence classification model is trained, a sample set including multiple types of conference samples needs to be obtained, generally, a manner of obtaining the sample set may be that a large amount of sample texts that are manually input form the sample set, or that a large amount of sample texts are read from other data obtaining devices or databases to form the sample set, and a manner of obtaining the sample set is specifically selected according to an actual situation, which is not limited in this embodiment of the present application.

It should be noted that each meeting sample in the sample set carries a type label, for example: the type label of the conference sample 'theme is environment protection' is 'theme type', and the type label of the conference sample 'after the conference, each member submits a planning report' is 'to-be-done type'.

Step S804: a first meeting sample is extracted from the sample set, wherein the first meeting sample is any meeting sample in the sample set.

Step S806: and inputting the first conference sample into a preset model to obtain a first prediction result of the first conference sample.

Step S808: and training the preset model according to the first prediction result and the type label carried by the first conference sample to obtain a trained sentence classification model.

Specifically, the preset model refers to a model that can classify sentences, including but not limited to a BERT model, and is specifically selected according to actual situations, which is not limited in this embodiment of the present application.

In one or more embodiments of the application, a first conference sample may be extracted from a sample set, where the first conference sample is any conference sample in the sample set, the first conference sample is input into a preset model to obtain a first prediction result of the first conference sample, and the preset model is trained according to the first prediction result and a type tag carried by the first conference sample to obtain a trained sentence classification model.

In practical application, there are various ways of training the preset model according to the first prediction result and the type tag carried by the first conference sample, and the method is specifically selected according to an actual situation, and the embodiment of the present application is not limited to this.

In a possible implementation manner, whether the current preset model is trained is determined by combining the iteration number.

Specifically, a first loss value is calculated according to a first prediction result and a type tag carried by a first conference sample, if the first loss value is larger than a first preset threshold value, model parameters of a preset model are adjusted, the step of extracting the first conference sample from the sample set is returned, the preset model is continuously trained, and iteration is stopped until the first preset iteration frequency is reached, so that a trained sentence classification model is obtained.

It should be noted that the first preset threshold and the first preset iteration number are selected according to actual situations, and this is not limited in this embodiment of the present application.

In practical applications, there are many functions for calculating the first loss value, such as a cross entropy loss function, an L1 norm loss function, a maximum loss function, a mean square error loss function, a logarithmic loss function, and the like.

Preferably, the cross entropy between the first prediction intent and the first intent tag carried by the first representation sample can be calculated as a first loss value using a cross entropy loss function, wherein the cross entropy loss function is as a common function

Shown in formula 1.

Wherein C represents the number of classes, p _i To true probability value, q _i To predict the probability value, i is the ith character.

By applying the scheme of the embodiment of the application, the cross entropy between the first prediction result and the type label carried by the first conference sample is calculated as the first loss value by utilizing the cross entropy loss function, so that the efficiency of calculating the first loss value is improved, and the efficiency of training the sentence classification model is improved.

In another possible implementation manner, the first loss value may be directly compared with a first preset threshold, and whether the current preset model is trained is determined.

Referring to fig. 9, fig. 9 shows a flowchart of training a sentence classification model in another method for extracting a key sentence in a meeting, which is provided in an embodiment of the present application, and specifically includes the following steps:

step S902: obtaining a sample set, wherein the sample set comprises a plurality of types of conference samples, and each conference sample carries a type label.

Step S904: a first meeting sample is extracted from the sample set, wherein the first meeting sample is any meeting sample in the sample set.

Step S906: and inputting the first conference sample into a preset model to obtain a first prediction result of the first conference sample.

Step S908: and calculating a first loss value according to the first prediction result and the type label carried by the first conference sample.

Step S910: and if the first loss value is larger than the first preset threshold value, adjusting the model parameters of the preset model, and returning to execute the step of extracting the first conference sample from the sample set.

Step S912: and if the first loss value is less than or equal to a first preset threshold value, stopping training to obtain a trained sentence classification model.

Specifically, if the first loss value is greater than the first preset threshold, it indicates that the difference between the first prediction result and the type tag carried by the first conference sample is large, the recognition capability of the trained sentence classification model for the first conference sample is poor, at this time, the model parameter of the trained sentence classification model can be adjusted, the step of extracting the first conference sample from the sample set is returned to be executed, the training of the preset model is continued until the first loss value is less than or equal to the first preset threshold, which indicates that the difference between the first prediction result and the type tag carried by the first conference sample is small, and the training is completed to obtain the trained sentence classification model.

It should be noted that the first preset threshold and the first preset iteration number are specifically selected according to actual situations, and this is not limited in this embodiment of the present application.

By applying the scheme of the embodiment of the application, the first prediction result and the first loss value of the type label carried by the first conference sample are calculated, the first loss value is compared with the first preset threshold value, the preset model is continuously trained under the condition that the first loss value is larger than the first preset threshold value until the training is completed under the condition that the first loss value is smaller than or equal to the first preset threshold value, the trained sentence classification model can be more accurate by continuously adjusting the model parameters of the preset model, and the accuracy of the sentence classification result is improved.

Referring to fig. 10, fig. 10 is a flowchart illustrating a seventh method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S1002: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

The expression of step S1002 is the same as that of step S202, and the description of this embodiment is omitted.

Step S1004: and matching the meeting text to be processed with the first target key sentence template to obtain a first meeting sentence corresponding to the target extraction type.

Step S1006: and matching the meeting text to be processed with a second target key sentence template to obtain a second meeting sentence corresponding to the target extraction type.

Step S1008: and inputting the conference text to be processed into a pre-trained sentence classification model, and acquiring a third conference sentence of the target extraction type.

Step S1010: and integrating the first conference statement, the second conference statement and the third conference statement to obtain a conference key statement corresponding to the target extraction type.

In one or more embodiments of the present application, after obtaining a to-be-processed conference text and a target extraction type of the to-be-processed conference text, in order to improve accuracy of a conference key sentence corresponding to the target extraction type in the to-be-processed conference text, a first target key sentence template, a second target key sentence template, and a sentence classification model may be respectively used to process the to-be-processed conference text, so as to obtain a first conference sentence, a second conference sentence, and a third conference sentence, and after obtaining the first conference sentence, the second conference sentence, and the third conference sentence, the first conference sentence, the second conference sentence, and the third conference sentence are integrated, so as to obtain a conference key sentence corresponding to the target extraction type in the to-be-processed conference text.

Specifically, the first target key sentence template refers to a key sentence template corresponding to the target extraction type of the conference text to be processed in a preset key sentence template library. The second target key sentence template is a key sentence template generated by replacing key contents in the preset dependency syntax template with the target extraction type. The pre-trained sentence classification model refers to a model which is pre-trained and can classify sentences.

Exemplarily, step S1004 is further explained: the method comprises the steps of obtaining a preset key sentence template library, wherein the preset key sentence template library comprises two theme type key sentence templates, and the two theme type key sentence templates respectively convey 'the notification of the meeting, the' theme is. The meeting text to be processed is' the meeting notice conveys the spirit of the education work and the key point content of the education work, the subject is caring for students and promoting the education development, and the conclusion of the meeting is that all units need to ask for and report the work to the leader in time, and the leader needs to work in a courage within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. The method comprises the steps of firstly, searching a first target key sentence template corresponding to a target extraction type of a to-be-processed conference text in a preset key sentence template base to obtain two first conference key sentences corresponding to the target extraction type of the to-be-processed conference text, wherein the two first conference key sentences are used for respectively conveying the spirit of education work and the key point content of the education work for the conference notification, and the theme is a concern for students to promote education development.

Further explanation is made to step S1006: the method comprises the steps of obtaining a preset dependency syntax template which is ' noun + be verb + clause ', obtaining a to-be-processed meeting text which is ' the meeting notice conveys the spirit of education work and the key point content of the education work, caring about students and promoting education development, and leading the leaders to ask and report the work to the leaders in time and to work in the scope of the job right. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. "the target extraction type of the conference text to be processed is" topic type ", the noun in the dependency syntax template is replaced by" topic ", the verb be is replaced by" yes ", the second target key sentence template is obtained as" topic yes ", the conference text to be processed is matched with the second target key sentence template, and the second conference key sentence" topic corresponding to the target extraction type is obtained, which is a concern for students and promotes education development ".

Further explanation is made to step S1008: the meeting text to be processed is acquired, namely that the meeting notice conveys the spirit of the education work and the key point content of the education work, the subject is to care students and promote the education development, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. The method comprises the steps of inputting a text of a meeting to be processed into a pre-trained sentence classification model, obtaining a meeting statement, wherein the sentence type of a meeting notice conveying the spirit of education work and the key point content of the education work is 'the type containing key information', the subject of the meeting statement is caring students and promoting education development 'is' the subject type ', the conclusion of the meeting statement' meeting is that each unit needs to ask for and report work to leaders in time, the sentence type of the leaders needs to work in a courage range is 'the conclusion type', the meeting statement 'is arranged in the following work, the diligence and economic tasks of students are well implemented, a teacher dining room is moved to the second floor, all first floor dining rooms are used by the students, and the sentence type of solving the problem of the crowding of the students' is 'the type to be done'. The target extraction type of the obtained conference text to be processed is a theme type, and the third conference sentence corresponding to the target extraction type of the obtained conference text to be processed is a theme caring student and promoting education development.

Integrating a first meeting statement, namely the meeting notice conveys the spirit of the education work and the key content of the education work, and a subject, namely caring for students and promoting education development, a subject, namely a caring student, promoting education development and a subject, namely a caring student and promoting education development, of a second meeting statement to obtain a meeting key sentence corresponding to the target extraction type.

In practical application, there are various methods for integrating the first conference statement, the second conference statement and the third conference statement to obtain the conference key statement corresponding to the target extraction type, and the method is specifically selected according to practical situations, and the embodiment of the present application does not limit this.

In a possible implementation manner, a first conference statement, a second conference statement and a third conference statement are integrated, and a conference statement with the largest occurrence frequency in the first conference statement, the second conference statement and the third conference statement is used as a conference key statement corresponding to a target extraction type of a conference text to be processed.

Referring to the above example, in the obtained first meeting statement, the second meeting statement and the third meeting statement, the meeting statement "the meeting notification conveys the spirit of the education work and the content of the main point of the education work" appears once, "the subject is a caring student and the education development is promoted" appears three times, and therefore, it is determined that the meeting key statement corresponding to the target extraction type "subject type" in the meeting text to be processed is the "subject is a caring student and the education development is promoted".

In another possible implementation manner, the first meeting statement, the second meeting statement and the third meeting statement are integrated, and the meeting statement in the first meeting statement, the second meeting statement and the third meeting statement, in which the number of words is greater than the preset number of words, is used as the meeting key statement corresponding to the target extraction type of the meeting text to be processed.

Referring to the above example, the preset word number is 14, in the obtained first, second and third meeting sentences, the number of words of the meeting sentence "the meeting notification conveys the spirit of the educational work and the content of the main point of the educational work" is 26, and the number of words of "the subject is a concern about the student and promotes the educational development" is 14, so that it is determined that the meeting key sentence corresponding to the target extraction type "subject type" in the meeting text to be processed is "the meeting notification conveys the spirit of the educational work and the content of the main point of the educational work".

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is matched with the first target key sentence template to obtain the first conference sentence corresponding to the target extraction type, the conference text to be processed is matched with the second target key sentence template to obtain the second conference sentence corresponding to the target extraction type, the conference text to be processed is input into a pre-trained sentence classification model to obtain the third conference sentence of the target extraction type, the candidate conference sentences of the conference key sentences are determined by using the three methods, the conference text to be processed is subjected to deep analysis, the first conference sentence, the second conference sentence and the third conference sentence are integrated to obtain the conference key sentence corresponding to the target extraction type, and the accuracy of the conference key sentence is improved.

Referring to fig. 11, fig. 11 is a flowchart illustrating an eighth method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S1102: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

Step S1104: and matching the meeting text to be processed with the first target key sentence template to obtain a first meeting sentence corresponding to the target extraction type.

Step S1106: and matching the meeting text to be processed with a second target key sentence template to obtain a second meeting sentence corresponding to the target extraction type.

Step S1108: and inputting the conference text to be processed into a pre-trained sentence classification model, and acquiring a third conference sentence of the target extraction type.

The expressions of steps S1102, S1104, S1106, and S1108 are the same as those of steps S1002, S1004, S1006, and S1008, and thus the description thereof is omitted.

Step S1110: and integrating the conference sentences of the same type in the first conference sentence, the second conference sentence and the third conference sentence.

Step S1112: and acquiring a plurality of conference sentences corresponding to the target extraction type, and taking the conference sentences meeting preset conditions as conference key sentences corresponding to the target extraction type.

In one or more embodiments of the application, after a target extraction type of a to-be-processed conference text and a to-be-processed conference text are obtained, the to-be-processed conference text is matched with a first target key sentence template to obtain a first conference sentence corresponding to the target extraction type, the to-be-processed conference text is matched with a second target key sentence template to obtain a second conference sentence corresponding to the target extraction type, the to-be-processed conference text is input into a pre-trained sentence classification model, a third conference sentence of the target extraction type is obtained, the same conference sentences in the first conference sentence, the second conference sentence and the third conference sentence can be integrated to obtain a plurality of conference sentences corresponding to the target extraction type, and the conference sentences meeting preset conditions are used as conference key sentences corresponding to the target extraction type.

Specifically, the preset condition refers to a preset condition for screening out a conference key sentence from a plurality of conference sentences, and is specifically selected according to an actual situation, which is not limited in this embodiment of the present application.

It should be noted that, a manner of integrating the conference sentences of the same type in the first conference sentence, the second conference sentence, and the third conference sentence may be to directly merge the conference sentences of the same type. Further, since the integrated conference statement may include a plurality of identical conference statements, a union set of conference statements of the same type in the first conference statement, the second conference statement, and the third conference statement may be taken, thereby improving user experience.

Illustratively, the meeting text to be processed is acquired as that the meeting notification conveys the spirit of the education work and the key point content of the education work, the subject is to care students and promote the education development, the conclusion of the meeting is that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a large amount within the scope of the authority. The conference arranges the following work, realizes the diligence and frugal tasks of students, moves the teacher's dining room to the second floor, and completely provides the students with the first floor dining room, so as to solve the problem of crowded dining of the students. "the target extraction types of the conference text to be processed are" topic type "and" conclusion type ".

The method comprises the steps of obtaining a preset key sentence template library, wherein the preset key sentence template library comprises two theme type key sentence templates, and the two theme type key sentence templates respectively convey 'the notification of the meeting, the' theme is. In a preset key sentence template base, searching a first target key sentence template corresponding to a target extraction type 'theme type' for 'the meeting notification conveys the meeting work' and 'the theme is.', matching a meeting text to be processed with the first target key sentence template to obtain two first meeting key sentences corresponding to the target extraction type, respectively 'the meeting notification conveys the spirit of the education work and the main point content of the education work' and 'the theme is about students and promotes education development', searching a first target key sentence template corresponding to a target extraction type 'conclusion type' as 'the conclusion of the meeting', matching the meeting text to be processed with the first target key sentence template to obtain three first meeting key sentences corresponding to the target extraction type, respectively informing the meeting notification conveys the spirit of the education work and the main point content of the education work The subject is concerned about students and promotes education development, and the conclusion of the conference corresponding to the conclusion type is that each unit needs to ask for and report work to leaders in time, and the leaders need to work in a large amount within the scope of the authority.

The method comprises the steps of obtaining a preset dependency syntax template as a noun + a verb be + a clause, aiming at a target extraction type of a theme type, replacing the noun in the dependency syntax template with a theme, replacing the verb be with a yes, obtaining a second target key sentence template as a theme yes, replacing the noun in the dependency syntax template with a conclusion, replacing the verb be with a yes, obtaining a second target key sentence template as a conclusion yes, matching a conference text to be processed with the second target key sentence template, obtaining a theme corresponding to the second conference key sentence of the target extraction type as a theme type, wherein the theme is a caring student, and the conclusion of a conference corresponding to promote educational development and the conclusion of the conference corresponding to the conclusion type is that each unit needs to be shown to the leader in time, report work, lead to work in great capacity within the scope of job title ".

The method comprises the steps of inputting a text of a meeting to be processed into a pre-trained sentence classification model, obtaining a meeting statement that the meeting notice conveys the spirit of education work and the key point content of the education work, wherein the sentence type of the meeting statement is ' a type containing key information ', the subject of the meeting statement is a subject of caring students and promoting education development ' is ' a subject type ', the conclusion of the meeting statement ' is that each unit needs to ask for and report work to leaders in time, the sentence type of the leaders needs to work in a courage range is ' a conclusion type ', the meeting statement ' arranges the following work in the meeting, realizes the diligence and economic tasks of students, moves a teacher restaurant to the second floor, and completely provides for the students in the first floor restaurant so as to solve the problem that the students have a crowded meal, and the sentence type is ' a type to be done '. And obtaining a third meeting statement corresponding to a target extraction type ' theme type ' of the meeting text to be processed as ' the theme is that students are concerned and education development is promoted ', and ' the third meeting statement corresponding to a conclusion type ' is that all units need to ask for and report work to the leader in time and the leader needs to work in a large amount within the scope of the authority '.

And merging the same types of conference sentences in the first conference sentence, the second conference sentence and the third conference sentence to obtain a conference sentence with a conclusion type, namely that each unit needs to ask for and report work to a leader in time and the leader needs to work in a large amount within the scope of the authority. The conference sentence of "topic type" includes "this conference notification conveys the spirit of the education work and the content of the main point of the education work" and "the topic is to care for the students, promote the education development".

The preset condition is that the number of words is larger than 14, so that the meeting key sentence corresponding to the target extraction type ' theme type ' in the meeting text to be processed is ' the meeting notification conveys the spirit of the education work and the key content of the education work ', and the meeting key sentence corresponding to the conclusion type ' is ' the conclusion that each unit needs to ask for and report the work to the leader in time, and the leader needs to work in a courage within the scope of the authority '.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are taken, the conference text to be processed is matched with the first target key sentence template to obtain the first conference sentence corresponding to the target extraction type, the conference text to be processed is matched with the second target key sentence template to obtain the second conference sentence corresponding to the target extraction type, the conference text to be processed is input into a pre-trained sentence classification model, and acquiring a third conference sentence of a target extraction type, integrating the conference sentences of the same type in the first conference sentence, the second conference sentence and the third conference sentence to acquire a plurality of conference sentences corresponding to the target extraction type, and taking the conference sentences meeting preset conditions as conference key sentences corresponding to the target extraction type, thereby further improving the accuracy of the conference key sentences.

Referring to fig. 12, fig. 12 is a flowchart illustrating a ninth method for extracting a meeting key sentence according to an embodiment of the present application, which may specifically include the following steps:

step S1202: and acquiring the conference text to be processed and the target extraction type of the conference text to be processed.

Step S1204: and matching the meeting text to be processed with the first target key sentence template to obtain a first meeting sentence corresponding to the target extraction type.

Step S1206: and matching the meeting text to be processed with a second target key sentence template to obtain a second meeting sentence corresponding to the target extraction type.

Step S1208: and inputting the conference text to be processed into a pre-trained sentence classification model, and acquiring a third conference sentence of the target extraction type.

The expressions of steps S1202, S1204, S1206, and S1208 are the same as those of steps S1002, S1004, S1006, and S1008, and the description thereof is omitted here.

Step 1210: and arranging the first conference statement, the second conference statement and the third conference statement according to a preset sequence to obtain an arranged conference statement sequence.

Step S1212: and acquiring a conference key sentence corresponding to the target extraction type according to the conference sentence sequence.

In one or more embodiments of the application, after a target extraction type of a to-be-processed conference text and a to-be-processed conference text are obtained, the to-be-processed conference text is matched with a first target key sentence template to obtain a first conference sentence corresponding to the target extraction type, the to-be-processed conference text is matched with a second target key sentence template to obtain a second conference sentence corresponding to the target extraction type, the processed conference text is input into a pre-trained sentence classification model, a third conference sentence of the target extraction type is obtained, the first conference sentence, the second conference sentence and the third conference sentence can be arranged according to a preset sequence to obtain an arranged conference sentence sequence, and a conference key sentence corresponding to the target extraction type is obtained according to the conference sentence sequence.

Specifically, the preset sequence refers to a preset condition for arranging the conference sentences, for example, an arrangement sequence is determined based on the number of words of the conference sentences, the conference sentences are arranged from a small number to a large number, or the conference sentences are arranged from a large number to a small number according to the number of words, and the like.

It should be noted that after the arranged conference sentence sequence is obtained, any conference sentence in the conference sentence sequence may be randomly selected as the conference key sentence corresponding to the target extraction type, or the conference key sentence corresponding to the target extraction type may be determined according to the word number of the conference sentence, for example, the conference sentence with the largest word number in the conference sentence sequence is used as the conference key sentence corresponding to the target extraction type, and of course, the conference key sentence corresponding to the target extraction type may also be selected from the conference sentence sequence according to other selection conditions, and the selection is specifically performed according to an actual situation, which is not limited in this embodiment of the present application.

Illustratively, referring to the example in fig. 11 above, the first meeting statement of "topic type" is obtained as "the meeting notification conveys the spirit of the education task and the content of the main point of the education task" and "the topic is that the students are concerned and the education development is promoted," the first meeting statement of "conclusion type" is that the units want to ask for and report the task to the leader in time, and the leader wants to work in a large amount within the scope of the job title, "respectively. The second meeting statement of the ' topic type ' is ' the topic is that students are concerned and education development is promoted ', and the ' conclusion type ' of the second meeting statement is that all units need to ask for and report work to leaders in time and leaders need to work in a large amount within the scope of the authority '. The third meeting statement of the 'topic type' is 'the topic is that students are concerned and education development is promoted', and the 'conclusion type' of the third meeting statement is 'the conclusion that each unit needs to ask for and report work to the leader in time and the leader needs to work in a large amount within the scope of the authority'.

Arranging a first conference statement, a second conference statement and a third conference statement according to a preset sequence of ' the first conference statement, the second conference statement and the third conference statement, wherein all the conference statements are arranged according to a preset sequence of which the number of words is from few to many ', and obtaining a ' theme type ' conference statement sequence as a ' theme is about students so as to promote educational development; the meeting notice conveys the spirit of the education work and the key point content of the education work; the theme is caring for students and promoting educational development; the subject is to care students and promote education development, and the conference sentence sequence of the conclusion type is the conclusion of the conference, namely, each unit needs to ask for and report work to leaders in time, and the leaders need to work in a large amount within the scope of the authority; the conference conclusion is that each unit needs to ask for and report work to the leader in time, and the leader needs to work in a large amount within the scope of the job right; the conference conclusion is that each unit needs to ask for and report work to the leader in time, the leader needs to work in a large amount within the scope of the authority, a conference key sentence corresponding to the target extraction type is randomly selected from the conference sentence sequence of the theme type and the conference sentence sequence of the conclusion type, the conference key sentence corresponding to the theme type is determined as that the conference notification conveys the spirit of the education work and the content of the key points of the education work, the conference key sentence of the conclusion type is that each unit needs to ask for and report work to the leader in time, and the leader needs to work in a large amount within the scope of the authority.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, the conference text to be processed is matched with the first target key sentence template to obtain the first conference sentence corresponding to the target extraction type, the conference text to be processed is matched with the second target key sentence template to obtain the second conference sentence corresponding to the target extraction type, the conference text to be processed is input into a pre-trained sentence classification model, acquiring a third conference statement of a target extraction type, arranging the first conference statement, the second conference statement and the third conference statement according to a preset sequence to obtain an arranged conference statement sequence, and acquiring the meeting key sentence corresponding to the target extraction type according to the meeting sentence sequence, so that the meeting key sentence can be acquired according to the requirement, and the accuracy of the meeting key sentence is improved.

It should be noted that the technical solution of the meeting key sentence extraction method belongs to the same concept as the technical solution of the meeting key sentence extraction method shown in fig. 1 to 12, and details of the technical solution of the meeting key sentence extraction method, which are not described in detail, can be referred to the description of the technical solution of the meeting key sentence extraction method shown in fig. 1 to 12.

Fig. 13 shows a schematic diagram of a method for extracting a meeting key sentence according to an embodiment of the present application, which specifically includes:

acquiring a conference text to be processed and a target extraction type of the conference text to be processed, and performing character string fuzzy matching on the conference text to be processed and a subject sentence template, a conclusion sentence template and a sentence template to be made respectively to obtain a first specified type key sentence; respectively carrying out dependency syntax matching on the conference text to be processed and the theme sentence pattern template, the conclusion sentence pattern template and the sentence pattern template to be made to obtain a second specified type key sentence; inputting the meeting text to be processed into a pre-trained sentence classification template, obtaining a third specified type key sentence and an important information key sentence, integrating the obtained first specified type key sentence, the second specified type key sentence, the third specified type key sentence and the important information key sentence, and obtaining the meeting key sentence of the meeting text to be processed.

By applying the scheme of the embodiment of the application, the conference text to be processed and the target extraction type of the conference text to be processed are obtained, and the conference text to be processed is subjected to character string fuzzy matching with a subject sentence template, a conclusion sentence template and a sentence template to be made respectively to obtain a first key sentence of a specified type; respectively carrying out dependency syntax matching on the conference text to be processed and the theme sentence pattern template, the conclusion sentence pattern template and the sentence pattern template to be made to obtain a second specified type key sentence; the method comprises the steps of inputting a conference text to be processed into a pre-trained sentence classification template, obtaining a third specified type key sentence and an important information key sentence, integrating the obtained first specified type key sentence, the second specified type key sentence, the third specified type key sentence and the important information key sentence, and obtaining the conference key sentence of the conference text to be processed.

Corresponding to the above method embodiment, the present application further provides an embodiment of a meeting key sentence extraction apparatus, and fig. 14 shows a schematic structural diagram of the meeting key sentence extraction apparatus provided in the embodiment of the present application. As shown in fig. 14, the apparatus includes:

an obtaining module 1402 configured to obtain a to-be-processed conference text and a target extraction type of the to-be-processed conference text;

the processing module 1404 is configured to identify the conference text to be processed according to the target extraction type, and obtain a conference key sentence corresponding to the target extraction type.

Optionally, the target extraction type includes at least one of a topic type, a conclusion type, a to-do type, and a key information containing type.

Optionally, the processing module 1404 is further configured to obtain a target key sentence template corresponding to the target extraction type according to the target extraction type;

and matching the meeting text to be processed with the target key sentence template to obtain a meeting key sentence corresponding to the target extraction type.

Optionally, the target key sentence template comprises a first target key sentence template;

the processing module 1404 is further configured to obtain a preset key sentence template base, where the key sentence template base includes multiple types of key sentence templates;

and searching a first target key sentence template corresponding to the target extraction type in a preset key sentence template base.

Optionally, the target key sentence template comprises a second target key sentence template;

a processing module 1404, further configured to obtain a preset dependency syntax template;

and performing keyword replacement on the dependency syntax template based on the target extraction type to obtain a second target key sentence template.

Optionally, the processing module 1404 is further configured to input the to-be-processed conference text into a pre-trained sentence classification model, and obtain a sentence type of each conference sentence in the to-be-processed conference text;

and acquiring the conference key sentence corresponding to the target extraction type based on the sentence type of each conference sentence.

Optionally, the processing module 1404 is further configured to classify each conference statement according to a sentence type;

and acquiring a plurality of conference sentences corresponding to the target extraction type, and taking the conference sentences meeting preset conditions as conference key sentences corresponding to the target extraction type.

Optionally, the processing module 1404 is further configured to match the meeting text to be processed with the first target key sentence template, so as to obtain a first meeting sentence corresponding to the target extraction type;

matching the meeting text to be processed with a second target key sentence template to obtain a second meeting statement corresponding to the target extraction type;

inputting the conference text to be processed into a pre-trained sentence classification model, and acquiring a third conference sentence of a target extraction type;

and integrating the first conference statement, the second conference statement and the third conference statement to obtain a conference key statement corresponding to the target extraction type.

Optionally, the processing module 1404 is further configured to integrate conference sentences of the same type in the first conference sentence, the second conference sentence, and the third conference sentence;

Optionally, the processing module 1404 is further configured to arrange the first conference statement, the second conference statement and the third conference statement in a preset order, so as to obtain an arranged conference statement sequence;

and acquiring a conference key sentence corresponding to the target extraction type according to the conference sentence sequence.

The above is a schematic scheme of a meeting key sentence extraction apparatus of this embodiment. It should be noted that the technical scheme of the meeting key sentence extraction apparatus and the technical scheme of the meeting key sentence extraction method belong to the same concept, and details of the technical scheme of the meeting key sentence extraction apparatus, which are not described in detail, can be referred to the description of the technical scheme of the meeting key sentence extraction method.

Further, the components in the device embodiment should be understood as functional blocks that must be created to implement the steps of the program flow or the steps of the method, and each functional block is not actually divided or separately defined. The device claims defined by such a set of functional modules are to be understood as a functional module framework for implementing the solution mainly by means of a computer program as described in the specification, and not as a physical device for implementing the solution mainly by means of hardware.

FIG. 15 illustrates a block diagram of a computing device 1500 provided in accordance with an embodiment of the present application. The components of the computing device 1500 include, but are not limited to, a memory 1510 and a processor 1520. The processor 1520 is coupled to the memory 1510 via a bus 1530 and a database 1550 is used to store data.

The computing device 1500 also includes an access device 1540 that enables the computing device 1500 to communicate via one or more networks 1560. Examples of such networks include a Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the internet. The Access device 1540 can include one or more of any type of Network Interface Card (e.g., a Network Interface Card) that is wired or Wireless, such as an IEEE802.11 Wireless Local Area Network (WLAN) Wireless Interface, a worldwide Interoperability for Microwave Access (Wi-MAX) Interface, an ethernet Interface, a Universal Serial Bus (USB) Interface, a cellular Network Interface, a bluetooth Interface, a Near Field Communication (NFC) Interface, and so forth.

In one embodiment of the application, the above-described components of computing device 1500, as well as other components not shown in FIG. 15, may also be connected to each other, such as by a bus. It should be understood that the block diagram of the computing device architecture shown in FIG. 15 is for purposes of example only and is not limiting as to the scope of the present application. Those skilled in the art may add or replace other components as desired.

Computing device 1500 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., tablet, personal digital assistant, laptop, notebook, netbook, etc.), mobile phone (e.g., smartphone), wearable computing device (e.g., smartwatch, smartglasses, etc.), or other type of mobile device, or a stationary computing device such as a desktop computer or PC. Computing device 1500 may also be a mobile or stationary server.

Among other things, processor 1520 is configured to execute the following computer-executable instructions:

The above is an illustrative scheme of a computing device of the present embodiment. It should be noted that the technical solution of the computing device and the technical solution of the above-mentioned meeting key sentence extracting method belong to the same concept, and details that are not described in detail in the technical solution of the computing device can be referred to the description of the technical solution of the above-mentioned meeting key sentence extracting method.

An embodiment of the present application further provides a computer readable storage medium storing computer instructions that, when executed by a processor, are configured to perform the steps of:

The above is an illustrative scheme of a computer-readable storage medium of the present embodiment. It should be noted that the technical solution of the storage medium and the technical solution of the above-mentioned meeting key sentence extracting method belong to the same concept, and details that are not described in detail in the technical solution of the storage medium can be referred to the description of the technical solution of the above-mentioned meeting key sentence extracting method.

The computer instructions comprise computer program code which may be in the form of source code, object code, an executable file or some intermediate form, or the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, etc.

An embodiment of the present application further provides a chip, in which a computer program is stored, and when executed by the chip, the computer program performs the following steps:

acquiring a meeting text to be processed and a target extraction type of the meeting text to be processed;

It should be noted that, for the sake of simplicity, the above-mentioned method embodiments are described as a series of acts or combinations, but those skilled in the art should understand that the present application is not limited by the described order of acts, as some steps may be performed in other orders or simultaneously according to the present application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

In the foregoing embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to the related descriptions of other embodiments.

The preferred embodiments of the present application disclosed above are intended only to aid in the explanation of the application. Alternative embodiments are not exhaustive and do not limit the invention to the precise embodiments described. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the application and its practical applications, to thereby enable others skilled in the art to best understand and utilize the application. The application is limited only by the claims and their full scope and equivalents.

Claims

1. A conference key sentence extraction method is characterized by comprising the following steps:

and identifying the meeting text to be processed according to the target extraction type to obtain a meeting key sentence corresponding to the target extraction type.

2. The method of claim 1, wherein the target extraction type comprises at least one of a topic type, a conclusion type, a to-do type, and a key information containing type.

3. The method according to claim 1, wherein the step of identifying the to-be-processed meeting text according to the target extraction type to obtain a meeting key sentence corresponding to the target extraction type includes:

obtaining a target key sentence template corresponding to the target extraction type according to the target extraction type;

4. The method of claim 3, wherein the target key sentence template comprises a first target key sentence template;

the step of obtaining a target key sentence template corresponding to the target extraction type according to the target extraction type comprises the following steps:

acquiring a preset key sentence template base, wherein the key sentence template base comprises a plurality of types of key sentence templates;

and searching a first target key sentence template corresponding to the target extraction type in the preset key sentence template base.

5. The method of claim 3, wherein the target key sentence template comprises a second target key sentence template;

the step of obtaining a target key sentence template corresponding to the target extraction type according to the target extraction type includes:

acquiring a preset dependency syntax template;

and performing keyword replacement on the dependency syntax template based on the target extraction type to obtain the second target key sentence template.

6. The method according to claim 1, wherein the step of identifying the to-be-processed meeting text according to the target extraction type to obtain a meeting key sentence corresponding to the target extraction type includes:

inputting the to-be-processed conference text into a pre-trained sentence classification model to obtain sentence types of all conference sentences in the to-be-processed conference text;

7. The method according to claim 6, wherein the step of obtaining the meeting key sentence corresponding to the target extraction type based on the sentence type of each meeting sentence comprises:

classifying the conference sentences according to the sentence types;

and acquiring a plurality of conference sentences corresponding to the target extraction type, and taking the conference sentences meeting preset conditions as the conference key sentences corresponding to the target extraction type.

8. The method according to claim 1, wherein the step of identifying the to-be-processed meeting text according to the target extraction type to obtain a meeting key sentence corresponding to the target extraction type includes:

matching the meeting text to be processed with a first target key sentence template to obtain a first meeting sentence corresponding to the target extraction type;

matching the meeting text to be processed with a second target key sentence template to obtain a second meeting sentence corresponding to the target extraction type;

inputting the conference text to be processed into a pre-trained sentence classification model, and acquiring a third conference sentence of the target extraction type;

and integrating the first conference statement, the second conference statement and the third conference statement to obtain a conference key sentence corresponding to the target extraction type.

9. The method of claim 8, wherein the step of integrating the first conference statement, the second conference statement and the third conference statement to obtain the conference key sentence corresponding to the target extraction type comprises:

integrating the same type of conference sentences in the first conference sentence, the second conference sentence and the third conference sentence;

10. The method of claim 8, wherein the step of integrating the first conference statement, the second conference statement and the third conference statement to obtain the conference key sentence corresponding to the target extraction type comprises:

arranging the first conference statement, the second conference statement and the third conference statement according to a preset sequence to obtain an arranged conference statement sequence;

and acquiring a meeting key sentence corresponding to the target extraction type according to the meeting sentence sequence.

11. A meeting key sentence extraction device is characterized by comprising:

the acquisition module is configured to acquire a to-be-processed conference text and a target extraction type of the to-be-processed conference text;

and the processing module is configured to identify the conference text to be processed according to the target extraction type to obtain a conference key sentence corresponding to the target extraction type.

12. A computing device, comprising:

a memory and a processor;

the memory is used for storing computer-executable instructions, and the processor is used for executing the computer-executable instructions to realize the steps of the conference key sentence extraction method of any one of claims 1 to 10.

13. A computer-readable storage medium storing computer instructions, which when executed by a processor implement the steps of the method for extracting a meeting key sentence according to any one of claims 1 to 10.