Method, device, electronic device and readable storage medium for determining enterprise events

Abstract

The present application provides a method, device, electronic device and readable storage medium for determining enterprise events, which obtain multiple enterprise event information of an enterprise to be evaluated, divide the multiple enterprise event information according to word meanings to determine multiple sub-event units; input each sub-event unit into a pre-trained event extraction model, and output multiple first event information; wherein the first event information is determined based on target annotation label information of the sub-event unit; input each sub-event unit into a pre-trained entity recognition model, and output multiple second event information; wherein the second event information is determined based on semantic rules in the entity recognition model; based on the multiple first event information and the multiple second event information, multiple reference event information are determined; based on the determined event types corresponding to each of the reference event information and the preset representative event types, at least one representative event of the enterprise to be evaluated is determined.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, electronic device and readable storage medium for determining enterprise events. Background Art

[0002] With the development of technology, when people want to know about a company's representative events, most of them rely on introductions from friends or search for the company's event information on the Internet to learn about the company's representative events. However, due to the excessive amount of company event information, it is impossible to extract the event information quickly and accurately, which leads to the inability to quickly and accurately understand the company's representative events.

[0003] At present, in order to quickly understand the detailed information of an enterprise and then conduct events on the enterprise, it is necessary to filter the event information of the enterprise, usually by dependency syntactic analysis. However, if there is an error in the dependency analysis in the noun phrase structure, it will directly lead to an error in the extraction of the entire noun phrase, thus affecting the result of event extraction. Therefore, how to quickly extract the important event information of the enterprise from the enterprise event information and then understand the representative events of the enterprise has become an urgent problem to be solved. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide a method, device, electronic device and readable storage medium for determining enterprise events, by semantically segmenting the event information of the enterprise to obtain multiple sub-event units, and simultaneously inputting the multiple sub-event units into the event extraction model and the entity recognition model, in the event extraction model, using the annotated label information of the sub-event unit to quickly determine the first event information, in the entity recognition model, using semantic rules to determine the second event information, using the first event information and the second event information to quickly and accurately determine the reference event information, and using the reference event information to quickly and accurately determine the representative event of the enterprise, thereby improving the efficiency of understanding the relevant information of the enterprise.

[0005] The present application embodiment provides a method for determining an enterprise event, the method comprising:

[0006] Acquire multiple enterprise event information of the enterprise to be evaluated, and divide the multiple enterprise event information into multiple sub-event units according to word meanings;

[0007] Input each of the sub-event units into a pre-trained event extraction model, and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit;

[0008] Input each of the sub-event units into a pre-trained entity recognition model, and output a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model;

[0009] Determine a plurality of reference event information based on a plurality of the first event information and a plurality of the second event information;

[0010] Based on the determined event types corresponding to the respective reference event information and the preset representative event types, at least one representative event of the enterprise to be evaluated is determined.

[0011] Further, the first event information is determined by the following steps:

[0012] Inputting the plurality of sub-event units into the event extraction model to determine a character string sequence corresponding to each sub-event unit;

[0013] For each of the sub-event units, based on the character string sequence of the sub-event unit, determine a target annotation label of the sub-event unit;

[0014] For one of the sub-event units, first event information corresponding to the sub-event unit is determined based on the target annotation label of the sub-event unit.

[0015] Furthermore, for each of the sub-event units, based on the character string sequence of the sub-event unit, determining a target annotation label of the sub-event unit includes:

[0016] Determining a plurality of reference annotation tags of the sub-event unit based on the character string sequence of the sub-event unit;

[0017] Scoring the multiple reference annotation labels to determine the reference annotation label corresponding to the highest label score among the multiple reference annotation labels;

[0018] The reference annotation label corresponding to the highest label score is used as the target annotation label for the sub-event unit.

[0019] Furthermore, the determining, based on the target annotation label of the sub-event unit, the first event information corresponding to the sub-event unit includes:

[0020] Get the preset event information custom template;

[0021] The event information corresponding to the target annotation label of the sub-event unit in the event information custom template is used to determine the first event information corresponding to the sub-event unit.

[0022] Further, the second event information is determined by the following steps:

[0023] Inputting the plurality of sub-event units into the entity recognition model to determine a plurality of named entity information, wherein the named entity information is composed of at least two of the sub-event units;

[0024] For each of the named entity information, based on a preset semantic rule, determining at least one trigger word of the named entity information within a preset character range;

[0025] For each of the named entity information, second event information corresponding to the named entity information is determined based on at least one trigger word of the named entity information and the named entity information.

[0026] Further, the determining of multiple reference event information based on multiple first event information and multiple second event information includes:

[0027] Event information deduplication processing is performed based on the plurality of first event information and the plurality of second event information, and the plurality of first event information and the plurality of second event information after the event information deduplication processing is completed are determined as a plurality of reference event information.

[0028] The embodiment of the present application also provides a device for determining an enterprise event, the device comprising:

[0029] An acquisition module, used to acquire multiple enterprise event information of the enterprise to be evaluated, and to segment the multiple enterprise event information according to word meanings to determine multiple sub-event units;

[0030] A first event information determination module, used to input each of the sub-event units into a pre-trained event extraction model, and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit;

[0031] A second event information determination module, used to input each of the sub-event units into a pre-trained entity recognition model, and output a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model;

[0032] A reference event information determination module, configured to determine a plurality of reference event information based on a plurality of the first event information and a plurality of the second event information;

[0033] The representative event determination module is used to determine at least one representative event of the enterprise to be evaluated based on the determined event types corresponding to the respective reference event information and the preset representative event types.

[0034] Further, the first event information determination module determines the first event information through the following steps:

[0035] Inputting the plurality of sub-event units into the event extraction model to determine a character string sequence corresponding to each sub-event unit;

[0036] For each of the sub-event units, based on the character string sequence of the sub-event unit, determine a target annotation label of the sub-event unit;

[0037] For one of the sub-event units, first event information corresponding to the sub-event unit is determined based on the target annotation label of the sub-event unit.

[0038] An embodiment of the present application also provides an electronic device, including: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the steps of the method for determining enterprise events as described above are performed.

[0039] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for determining an enterprise event as described above are executed.

[0040] The present application provides a method for determining an enterprise event, the method comprising: obtaining multiple enterprise event information of an enterprise to be evaluated, dividing the multiple enterprise event information according to word meaning to determine multiple sub-event units; inputting each of the sub-event units into a pre-trained event extraction model, and outputting multiple first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit; inputting each of the sub-event units into a pre-trained entity recognition model, and outputting multiple second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model; based on multiple first event information and multiple second event information, determining multiple reference event information; based on the event types corresponding to each of the determined reference event information and the preset representative event types, determining at least one representative event of the enterprise to be evaluated.

[0041] In this way, multiple sub-event units are obtained by semantically segmenting the enterprise's event information, and the multiple sub-event units are simultaneously input into the event extraction model and the entity recognition model. The first event information is quickly determined in the event extraction model using the annotation label information of the sub-event unit, and the second event information is determined by semantic rules in the entity recognition model. The reference event information is quickly and accurately determined using the first event information and the second event information, and the representative event of the enterprise is quickly and accurately determined using the reference event information, thereby improving the efficiency of understanding enterprise-related information.

[0042] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 A flowchart of a method for determining an enterprise event provided in an embodiment of the present application;

[0045] Figure 2 A flowchart of another method for determining an enterprise event provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of the structure of a device for determining an enterprise event provided in an embodiment of the present application;

[0047] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art, under the guidance of the content of the present application, can add one or more other operations to the flowchart, or remove one or more operations from the flowchart.

[0049] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0050] In order to enable those skilled in the art to use the contents of this application, the following implementation method is given in combination with the specific application scenario of "determining enterprise representative events". For those skilled in the art, the general principles defined here can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application.

[0051] The following methods, devices, electronic devices or computer-readable storage media of the embodiments of the present application can be applied to any scenario where enterprise representative events need to be determined. The embodiments of the present application are not limited to specific application scenarios. Any solution using a method, device, electronic device and computer-readable storage medium for determining an enterprise event provided by the embodiments of the present application is within the scope of protection of the present application.

[0052] First, the application scenarios to which the present application is applicable are introduced. The present application can be applied to the field of data processing technology.

[0053] With the development of technology, when people want to know about a company's representative events, most of them rely on introductions from friends or search for the company's event information on the Internet to learn about the company's representative events. However, due to the excessive amount of company event information, it is impossible to extract the event information quickly and accurately, which leads to the inability to quickly and accurately understand the company's representative events.

[0054] According to research, at present, in order to quickly understand the detailed information of an enterprise and then conduct events on the enterprise, it is necessary to filter the event information of the enterprise, usually by dependency syntactic analysis. However, if there is an error in the dependency analysis in the noun phrase structure, it will directly lead to an error in the extraction of the entire noun phrase, thus affecting the result of event extraction. Therefore, how to quickly extract the important event information of the enterprise from the enterprise event information and then understand the representative events of the enterprise has become an urgent problem to be solved.

[0055] There are currently the following methods for extracting event information from corporate public opinion information: Encode the original text and text annotation information. Input the result into a bidirectional LSTM (Long Short-Term Memory, LSTM). Obtain the forward and backward implicit state vectors with time sequence. First, parse the input sentence into a dependency tree structure, and then input the result of step 1 into the constructed dependency Tree-LSTM to obtain the implicit state vector of the root node and the implicit state vector at each moment. Obtain the feature vector that stores the entity relationship sentence information. At the same time, connect the forward and backward implicit state vectors of the bidirectional LSTM at time t and the implicit state vector of the dependency Tree-LSTM (TreeLong Short-Term Memory, Tree-LSTM) at time t. Perform trigger word recognition and classification; perform event argument recognition and classification. This technical solution uses dependency syntactic analysis. Since the dependency syntactic tree is based on the dependency relationship between words, if there is a dependency analysis error in the noun phrase structure, it will cause an error in the extraction of the entire noun phrase, thereby affecting the result of event extraction. Therefore, the accuracy requirement for the dependency analyzer is very high.

[0056] Based on this, the present application provides a method, device, electronic device and readable storage medium for determining enterprise events. The semantics of the enterprise's event information are segmented to obtain multiple sub-event units, and the multiple sub-event units are simultaneously input into an event extraction model and an entity recognition model. In the event extraction model, the first event information is quickly determined using the annotated label information of the sub-event unit, and in the entity recognition model, the second event information is determined by semantic rules. The first event information and the second event information are used to quickly and accurately determine the reference event information, and the reference event information is used to quickly and accurately determine the representative event of the enterprise, thereby improving the efficiency of understanding the enterprise-related information.

[0057] See also Figure 1 , Figure 1 This is a flow chart of a method for determining an enterprise event provided by an embodiment of the present application. Figure 1 As shown in , the determination method provided in the embodiment of the present application includes:

[0058] S101: Acquire multiple enterprise event information of the enterprise to be evaluated, and segment the multiple enterprise event information according to word meanings to determine multiple sub-event units.

[0059] In this step, multiple enterprise event information of the enterprise to be evaluated is obtained, and the multiple enterprise event information is segmented according to word meanings to determine multiple sub-event units.

[0060] Here, the enterprise event information refers to reports on the enterprise at some time or other types of information such as evaluations of the enterprise by some employees.

[0061] Here, the sub-event unit is obtained by segmenting the enterprise event information according to the word meaning. For example, after segmenting the word meaning of "executive Xiao Ming", the obtained "senior", "executive", "small" and "Ming" are all sub-event units.

[0062] Here, in an embodiment of the present application, an enterprise to be evaluated may be determined from multiple enterprises, and multiple enterprise event information of the enterprise to be evaluated may be obtained based on the name of the enterprise to be evaluated or related information. The enterprise event information of the enterprise may be obtained in a forum or in the media, and the source of the enterprise event information is not limited here.

[0063] Among them, the enterprise event information is segmented according to the meaning of words to determine multiple sub-event units. For example, "Most employees of the first company resigned" can be segmented into "first", "first", "company", "big", "many", "number", "employee", "worker", "resignation", "job" or "first company", "most", "employee", "resignation" according to the meaning of "most employees of the first company resigned". This part does not limit the method of segmenting enterprise event information by word meaning.

[0064] S102: Input each of the sub-event units into a pre-trained event extraction model, and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit.

[0065] In this step, each sub-event unit is input into the event extraction model to determine a plurality of first event information, wherein the first event information is determined based on the target annotation label information of the sub-event unit.

[0066] The event extraction model is trained in the following way. Obtain multiple enterprise event information, perform text sentence segmentation on the text content in the enterprise event information, and determine multiple target short sentences; label each sub-event unit in the target short sentence, and determine the sub-event unit carrying the sample label; wherein, for each sub-event unit, label the sub-event unit according to the event information of the sub-event unit; convert the multiple sub-event units carrying the sample label into a string sequence, and determine the string sequence of multiple sub-event units, wherein the encoding sequence of the sub-event unit carries the sample label information of the sub-event unit; input the string sequence of the multiple sub-event units into a pre-built classification model, and train to obtain the event extraction model. In this way, multiple first event information determined by the event extraction model is realized.

[0067] Here, the target annotation label information of the sub-event unit is the event information of the sub-event unit. For example, the annotation format of the sub-event unit is (event argument: event type-argument role), the event argument is "a senior executive of a company left", the event type is "left", and the argument role is "senior executive".

[0068] Further, the first event information is determined by the following steps:

[0069] A: Input the plurality of sub-event units into the event extraction model to determine the character string sequence corresponding to each sub-event unit.

[0070] Therein, a plurality of sub-event units are input into the event extraction model to determine the character string sequence corresponding to each sub-event unit.

[0071] Among them, multiple sub-event units are input into the event extraction model in the order of numbers. The BERT (Bidirectional Encoder Representations from Transformers, BERT) layer in the event model will convert the character form of each sub-event unit into a semantic coding sequence of characters, that is, a string sequence, according to certain conversion rules. The string sequence can be formed by the conversion relationship between the string and the character encoding.

[0072] B: For each of the sub-event units, based on the character string sequence of the sub-event unit, determine the target annotation label of the sub-event unit.

[0073] Wherein, the target annotation label of the sub-event unit is determined according to the character string sequence of the sub-event unit.

[0074] Here, for example, among the multiple sub-event units "a certain company executive leaves", according to the character string sequence "102354" of "a certain", it is determined that the target annotation label carried by the sub-event unit "a certain" is "a certain company, leaves, executive".

[0075] Further, based on the character string sequence of the sub-event unit, a target annotation label of the sub-event unit is determined, including:

[0076] a: Based on the character string sequence of the sub-event unit, determine multiple reference annotation tags of the sub-event unit.

[0077] Wherein, a plurality of reference annotation labels of the sub-event unit are determined according to the character string sequence of the sub-event unit.

[0078] Among them, because the same sub-event unit may appear many times in multiple enterprise event information, there will be multiple character string sequences about the sub-event unit, which will further lead to multiple reference annotation tags for the sub-event unit.

[0079] In a specific embodiment of the present application, the character string sequence of the sub-event unit "one" in the first enterprise event information is "1023", and the character string sequence of the sub-event unit "one" in the second enterprise event information is "1024". According to the character string sequence "1023" of the sub-event unit "one", the reference annotation label is determined to be "first company, resignation, employee", and according to the character string sequence "1024" of the sub-event unit "one", the reference annotation label is determined to be "first department, employee makes a mistake, employee".

[0080] b: Performing label scoring on the multiple reference label tags, and determining the reference label tag corresponding to the highest label score among the multiple reference label tags.

[0081] Wherein, label scoring is performed according to the determined multiple reference label labels, and a reference label label corresponding to the highest label score among the multiple reference label labels is determined.

[0082] Here, the reference annotation tags may be scored by content weight, and this part does not limit the way of scoring the reference annotation tags.

[0083] c: The reference annotation label corresponding to the highest label score is used as the target annotation label for the sub-event unit.

[0084] The reference annotation label corresponding to the highest label score is determined as the target annotation label of the sub-event unit.

[0085] C: For one of the sub-event units, based on the target annotation label of the sub-event unit, determine the first event information corresponding to the sub-event unit.

[0086] Wherein, the first event information corresponding to the sub-event unit is determined according to the target annotation label of the sub-event unit.

[0087] Here, the first event information is the event information corresponding to the target annotation label of the sub-event unit.

[0088] Furthermore, the determining, based on the target annotation label of the sub-event unit, the first event information corresponding to the sub-event unit includes:

[0089] 1): Get the preset event information custom template.

[0090] Among them, a preset event information custom template is obtained, where the event information custom template stores a large amount of event information corresponding to target annotation tags.

[0091] The event custom template stores event information corresponding to a large number of target labeling tags. The event information stored in the event custom template can be "First Company Bankruptcy", "First Company Executive Xiao Ming Was Dismissed" and other event information about the company. In addition, the event information in the event custom template corresponds to the target labeling tag one by one, and the corresponding event information can be quickly determined in the event custom template based on the target labeling tag number information.

[0092] The event information about the enterprise stored in the event custom template may be determined based on historical event information.

[0093] 2): The event information corresponding to the target annotation label of the sub-event unit in the event information custom template is used to determine the first event information corresponding to the sub-event unit.

[0094] The first event information corresponding to the sub-event unit is determined according to the event information corresponding to the target annotation tag of the sub-event unit in the event information custom template.

[0095] Among them, the numbering information of the target annotation label of the sub-event unit can be used to search in the event information custom template. When the numbering information of the target annotation label of the sub-event unit is successfully found, the corresponding event information is called out in the event information custom template according to the numbering information, and the event information is determined as the first event information corresponding to the sub-event unit.

[0096] In a specific embodiment of the present application, multiple sub-event units "first", "company", "senior", "executive", "out", and "leave" are input into the event extraction model, and the target annotation label information of each of the above sub-event units is determined in the event extraction model. For example, the target annotation label information of "one" is (first company, leave, senior executive), and then the first event information corresponding to the sub-event unit is determined as "first company senior executive leaves" in the event information custom template according to the numbering information of the target annotation label information. The determination process of the first event information corresponding to other sub-event units is consistent with the above determination process, and this part will not be repeated here.

[0097] S103: Input each of the sub-event units into a pre-trained entity recognition model, and output a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model.

[0098] In this step, each sub-event unit is input into the entity recognition model, and a plurality of second event information is output, wherein the second event information is determined according to the semantic rules in the entity recognition model.

[0099] Here, the entity recognition model is used to identify entity information in enterprise event information, such as "First Company", "Xiao Ming" and other entity information.

[0100] Here, the second event information is the event information corresponding to the sub-event unit.

[0101] Here, the semantic rules in the entity recognition model are the semantic rules between the first sub-event units, such as subject-predicate-object semantic rules and other forms of rules.

[0102] In a specific embodiment of the present application, multiple sub-event units "Xiao Ming", "First", "Company", "Executive", "Out", and "Leave" are input into the entity recognition model. In the entity recognition model, the entity information is first determined to be "Xiao Ming" and "Company", and then according to the semantic rules, it is determined that the entity information corresponds to the event information "First company executive Xiao Ming leaves". The event information "First company executive Xiao Ming leaves" is used as the second event information of the sub-event unit.

[0103] Further, the second event information is determined by the following steps:

[0104] I: Inputting a plurality of the sub-event units into the entity recognition model to determine a plurality of named entity information, wherein the named entity information is composed of at least two of the sub-event units.

[0105] Among them, multiple sub-event units are input into the entity recognition model to determine multiple named entity information.

[0106] Among them, the named entity information is generally a company name, a person's name, a department name, etc., and the named entity information is composed of at least two sub-event units, such as "Xiao Ming", "First Company", etc.

[0107] II: For each of the named entity information, based on a preset semantic rule, determine at least one trigger word of the named entity information within a preset character range.

[0108] Wherein, at least one trigger word within a preset character range is determined for each named entity information according to a preset semantic rule.

[0109] Among them, the trigger word is the verb that triggers the event, such as "leave" in "executives leave", "bankruptcy" in "the first company goes bankrupt", "layoffs" in "older employees are laid off" and other verbs.

[0110] Among them, the preset character range is between multiple characters of the named entity information, such as a certain character range before and after "company". The preset character range here can be set based on expert experience or using historical data. This part does not limit the setting method of the preset character range.

[0111] III: For each of the named entity information, based on at least one trigger word of the named entity information and the named entity information, determine the second event information corresponding to the named entity information.

[0112] Wherein, the second event information corresponding to the named entity information is determined according to the named entity information and at least one trigger word.

[0113] In a specific embodiment of the present application, "Xiao", "Ming", "Su", "In", "Of", "Department", "Gate", "Out", "Appear", "Big", "Quantity", "Employee", "Worker", "Resign", and "Job" are input into the entity recognition model, and the named entity information "Xiao Ming", "Department" and "Employee" are identified in the entity recognition model. According to the preset semantic rules, it is determined that the trigger word "resignation" appears within the 13-character range of "Xiao Ming". Then, according to the trigger word and the named entity, the second event information of "a large number of employees resigned from the department where Xiao Ming is located" is determined. Here, "department" or "employee" can also be used to determine that the corresponding second event information is also "a large number of employees resigned from the department where Xiao Ming is located", and the determination method is consistent with the above process, which will not be repeated in this part.

[0114] S104: Determine a plurality of reference event information based on a plurality of first event information and a plurality of second event information.

[0115] In this step, a plurality of reference event information is determined according to a plurality of first event information and a plurality of second event information.

[0116] Further, the determining of multiple reference event information based on multiple first event information and multiple second event information includes:

[0117] Event information deduplication processing is performed based on the plurality of first event information and the plurality of second event information, and the plurality of first event information and the plurality of second event information after the event information deduplication processing is completed are determined as a plurality of reference event information.

[0118] Among them, for the multiple reference event information and the multiple target event information, the reference event information and the target event information are judged, and if the reference time information is consistent with the target event information, the reference event information is deduplicated; if the reference time information is inconsistent with the target event information, the reference event information and the target event information are output.

[0119] In a specific embodiment of the present application, if it is determined that the first event information and the second event information have duplicate event information, then either the first event information or the second event information is deduplicated and only one event information is retained; if it is determined that the first event information and the second event information do not have duplicate event information, then the first event information and the second event information are determined together as reference event information.

[0120] S105: Determine at least one representative event of the enterprise to be evaluated based on the determined event types corresponding to the respective reference event information and preset representative event types.

[0121] In this step, at least one representative event of the enterprise to be evaluated is determined based on the event types corresponding to the determined reference event information and the preset representative event types. Here, the event types may include positive event types, negative event types, entertainment event types and other types.

[0122] Here, the preset representative event type may be a charity event type, a financing event type, or other types.

[0123] Here, the representative event may be a representative event in the positive event type. Here, the type of the representative event is not limited. For example, if the enterprise often mails supplies to poor mountainous areas, it can be determined that the representative event of the enterprise is charity.

[0124] Furthermore, the representative events of the enterprise to be evaluated may be determined in the following manner.

[0125] Based on the event types of multiple reference event information, representative events of the enterprise to be evaluated are determined, and the event types in the event types are compared with preset representative event types. If the event type of the enterprise belongs to the preset representative event type, at least one enterprise event in the event type is used as a representative event of the preset representative event type.

[0126] In the specific embodiments of this application, please refer to Figure 2 , Figure 2 This is a flow chart of another method for determining an enterprise event provided by an embodiment of the present application. Figure 2 As shown, multiple enterprise event information is input, multiple enterprise event information is segmented to determine multiple sub-event units, and each sub-event unit is simultaneously input into the event extraction model and the entity recognition model, and the first event information and the second event information are quickly and accurately determined, and then the reference event information is determined based on the first event information and the second event information, and then the level of the enterprise is quickly and accurately determined using the reference event information. For example, "small", "Ming", "where", "in", "of", "department", "door", "out", "now", "large", "quantity", "staff", "worker", "leave", "job" are simultaneously input into the event extraction model and the entity recognition model, and the target annotation label information of each sub-event unit is determined in the event extraction model according to the string sequence of multiple sub-event units, such as the target annotation label information of "small" is "leave, employee, department", and the first event information corresponding to the sub-event unit is determined according to the target annotation label information as "a large number of employees leave the department where Xiao Ming is located". Furthermore, in the entity recognition model, the named entity information "Xiao Ming", "department" and "employee" are identified in the entity recognition model. According to the preset semantic rules, the trigger word "resignation" appears within the 13 characters of "Xiao Ming". Then, according to the trigger word and the named entity, the second event information "a large number of employees resigned from the department where Xiao Ming works" is determined. After multiple first event information and multiple second event information are determined and deduplicated, reference event information is determined, and the representative event of the enterprise to be evaluated is determined using the event type of the reference event information.

[0127] The present application provides a method for determining an enterprise event, the method comprising: obtaining multiple enterprise event information of an enterprise to be evaluated, dividing the multiple enterprise event information according to word meaning to determine multiple sub-event units; inputting each of the sub-event units into a pre-trained event extraction model, and outputting multiple first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit; inputting each of the sub-event units into a pre-trained entity recognition model, and outputting multiple second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model; based on multiple first event information and multiple second event information, determining multiple reference event information; based on the event types corresponding to each of the determined reference event information and the preset representative event types, determining at least one representative event of the enterprise to be evaluated.

[0128] In this way, multiple sub-event units are obtained by semantically segmenting the enterprise's event information, and the multiple sub-event units are simultaneously input into the event extraction model and the entity recognition model. The first event information is quickly determined in the event extraction model using the annotation label information of the sub-event unit, and the second event information is determined by semantic rules in the entity recognition model. The reference event information is quickly and accurately determined using the first event information and the second event information, and the representative event of the enterprise is quickly and accurately determined using the reference event information, thereby improving the efficiency of understanding enterprise-related information.

[0129] See also Figure 3 , Figure 3 This is a schematic diagram of the structure of a device for determining an enterprise event provided in an embodiment of the present application. Figure 3 As shown in , the determining device 300 includes:

[0130] The acquisition module 310 is used to acquire multiple enterprise event information of the enterprise to be evaluated, and to segment the multiple enterprise event information according to word meanings to determine multiple sub-event units;

[0131] A first event information determination module 320 is used to input each of the sub-event units into a pre-trained event extraction model and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit;

[0132] A second event information determination module 330, for inputting each of the sub-event units into a pre-trained entity recognition model, and outputting a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model;

[0133] A reference event information determination module 340, configured to determine a plurality of reference event information based on a plurality of first event information and a plurality of second event information;

[0134] The representative event determination module 350 is used to determine at least one representative event of the enterprise to be evaluated based on the determined event types corresponding to the respective reference event information and the preset representative event types.

[0135] Further, the first event information determination module determines the first event information through the following steps:

[0136] Inputting the plurality of sub-event units into the event extraction model to determine a character string sequence corresponding to each sub-event unit;

[0137] For each of the sub-event units, based on the character string sequence of the sub-event unit, determine a target annotation label of the sub-event unit;

[0138] For one of the sub-event units, first event information corresponding to the sub-event unit is determined based on the target annotation label of the sub-event unit.

[0139] Further, when the first event information determination module 320 is used to determine the target annotation label of each sub-event unit based on the string sequence of the sub-event unit, the first event information determination module 320 is used to:

[0140] Determining a plurality of reference annotation tags of the sub-event unit based on the character string sequence of the sub-event unit;

[0141] Scoring the multiple reference annotation labels to determine the reference annotation label corresponding to the highest label score among the multiple reference annotation labels;

[0142] The reference annotation label corresponding to the highest label score is used as the target annotation label for the sub-event unit.

[0143] Furthermore, when the first event information determination module 320 is used to determine the first event information corresponding to the sub-event unit based on the target label tag of the sub-event unit, the first event information determination module 320 is also used to:

[0144] Get the preset event information custom template;

[0145] The event information corresponding to the target annotation label of the sub-event unit in the event information custom template is used to determine the first event information corresponding to the sub-event unit.

[0146] Further, the second event information determination module 330 determines the second event information through the following steps:

[0147] Inputting the plurality of sub-event units into the entity recognition model to determine a plurality of named entity information, wherein the named entity information is composed of at least two of the sub-event units;

[0148] For each of the named entity information, based on a preset semantic rule, determining at least one trigger word of the named entity information within a preset character range;

[0149] For each of the named entity information, second event information corresponding to the named entity information is determined based on at least one trigger word of the named entity information and the named entity information.

[0150] Further, when the reference event information determination module 340 is used to determine multiple reference event information based on multiple first event information and multiple second event information, the reference event information determination module 340 is also used to:

[0151] Event information deduplication processing is performed based on the plurality of first event information and the plurality of second event information, and the plurality of first event information and the plurality of second event information after the event information deduplication processing is completed are determined as a plurality of reference event information.

[0152] An embodiment of the present application provides a device for determining an enterprise event, the device comprising: an acquisition module, used to acquire multiple enterprise event information of an enterprise to be evaluated, and to determine multiple sub-event units by segmenting the multiple enterprise event information according to word meanings; a first event information determination module, used to input each of the sub-event units into a pre-trained event extraction model, and output multiple first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit; a second event information determination module, used to input each of the sub-event units into a pre-trained entity recognition model, and output multiple second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model; a reference event information determination module, used to determine multiple reference event information based on multiple first event information and multiple second event information; a representative event determination module, used to determine at least one representative event of the enterprise to be evaluated based on the event type corresponding to each of the determined reference event information and a preset representative event type.

[0153] In this way, multiple sub-event units are obtained by semantically segmenting the enterprise's event information, and the multiple sub-event units are simultaneously input into the event extraction model and the entity recognition model. The first event information is quickly determined in the event extraction model using the annotation label information of the sub-event unit, and the second event information is determined by semantic rules in the entity recognition model. The reference event information is quickly and accurately determined using the first event information and the second event information, and the representative event of the enterprise is quickly and accurately determined using the reference event information, thereby improving the efficiency of understanding enterprise-related information.

[0154] See also Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 4 As shown in , the electronic device 400 includes a processor 410 , a memory 420 and a bus 430 .

[0155] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 communicates with the memory 420 via the bus 430. When the machine-readable instructions are executed by the processor 410, the above-mentioned Figure 1 as well as Figure 2 The steps of a method for determining an enterprise event in the method embodiment shown, and the specific implementation method can be found in the method embodiment, which will not be repeated here.

[0156] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 1 as well as Figure 2 The steps of a method for determining an enterprise event in the method embodiment shown, and the specific implementation method can be found in the method embodiment, which will not be repeated here.

[0157] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0158] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0159] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0160] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0161] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0162] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for determining an enterprise event, characterized in that: The determination method comprises: Acquire multiple enterprise event information of the enterprise to be evaluated, and divide the multiple enterprise event information into multiple sub-event units according to word meanings; Input each of the sub-event units into a pre-trained event extraction model, and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit; Input each of the sub-event units into a pre-trained entity recognition model, and output a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model; Determine a plurality of reference event information based on a plurality of the first event information and a plurality of the second event information; Determine at least one representative event of the enterprise to be evaluated based on the determined event types corresponding to the respective reference event information and the preset representative event types; The first event information is determined by the following steps: Inputting the plurality of sub-event units into the event extraction model to determine a character string sequence corresponding to each sub-event unit; For each of the sub-event units, based on the character string sequence of the sub-event unit, determine a target annotation label of the sub-event unit; For one of the sub-event units, first event information corresponding to the sub-event unit is determined based on the target annotation label of the sub-event unit.

2. The determination method according to claim 1, characterized in that: The step of determining, for each of the sub-event units, a target annotation label of the sub-event unit based on the character string sequence of the sub-event unit includes: Determining a plurality of reference annotation tags of the sub-event unit based on the character string sequence of the sub-event unit; Scoring the multiple reference annotation labels to determine the reference annotation label corresponding to the highest label score among the multiple reference annotation labels; The reference annotation label corresponding to the highest label score is used as the target annotation label for the sub-event unit.

3. The determination method according to claim 1, characterized in that: The determining, based on the target annotation label of the sub-event unit, the first event information corresponding to the sub-event unit includes: Get the preset event information custom template; The event information corresponding to the target annotation label of the sub-event unit in the event information custom template is used to determine the first event information corresponding to the sub-event unit.

4. The determination method according to claim 1, characterized in that: The second event information is determined by the following steps: Inputting the plurality of sub-event units into the entity recognition model to determine a plurality of named entity information, wherein the named entity information is composed of at least two of the sub-event units; For each of the named entity information, based on a preset semantic rule, determining at least one trigger word of the named entity information within a preset character range; For each of the named entity information, second event information corresponding to the named entity information is determined based on at least one trigger word of the named entity information and the named entity information.

5. The determination method according to claim 1, characterized in that: The determining of a plurality of reference event information based on the plurality of first event information and the plurality of second event information comprises: Event information deduplication processing is performed based on the plurality of first event information and the plurality of second event information, and the plurality of first event information and the plurality of second event information after the event information deduplication processing is completed are determined as a plurality of reference event information.

6. A device for determining an enterprise event, characterized in that: The determining device comprises: An acquisition module, used to acquire multiple enterprise event information of the enterprise to be evaluated, and to segment the multiple enterprise event information according to word meanings to determine multiple sub-event units; A first event information determination module, used to input each of the sub-event units into a pre-trained event extraction model, and output a plurality of first event information; wherein the first event information is determined based on the target annotation label information of the sub-event unit; A second event information determination module, used to input each of the sub-event units into a pre-trained entity recognition model, and output a plurality of second event information; wherein the second event information is determined based on the semantic rules in the entity recognition model; A reference event information determination module, configured to determine a plurality of reference event information based on a plurality of the first event information and a plurality of the second event information; A representative event determination module, configured to determine at least one representative event of the enterprise to be evaluated based on the determined event types corresponding to the respective reference event information and the preset representative event types; The first event information determination module determines the first event information through the following steps: Inputting the plurality of sub-event units into the event extraction model to determine a character string sequence corresponding to each sub-event unit; For each of the sub-event units, based on the character string sequence of the sub-event unit, determine a target annotation label of the sub-event unit; For one of the sub-event units, first event information corresponding to the sub-event unit is determined based on the target annotation label of the sub-event unit.

7. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to perform the steps of a method for determining an enterprise event as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for determining an enterprise event as claimed in any one of claims 1 to 5 are executed.

Images