Document-Level Event Argument Extraction Method Based on Role-Focused Heterogeneous Graph

Abstract

The present invention proposes a document-level event argument extraction method based on a role-focus heterogeneous graph. The method includes: associating event types with role nodes; associating trigger words and central sentences in an event instance to construct a role-focus heterogeneous graph; inputting the role-focus heterogeneous graph into a graph attention network for learning to obtain node representations after learning by the graph attention network; using a pre-trained language model to capture document words from the node representations after learning by the graph attention network and performing weighted fusion with the node representations after learning by the graph attention network; the present invention develops a document-level event argument extraction model based on a role-focus heterogeneous graph. By using a graph neural network, role representations updated through the role-focus heterogeneous graph are obtained, and an argument span serving as the role is obtained by adopting a role-based span selection strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of information extraction, and particularly to a document-level event argument extraction method based on a role-focus heterogeneous graph. Background Art

[0002] Since existing research on document-level event recognition and its argument extraction all deal with the case of entity arguments, in order to expand the applicable scope of event extraction, a large number of scholars have focused on the research of document-level event argument extraction for complex arguments. A complex argument refers to a complex object described by a piece of text serving as an event argument, which may be a text fragment composed of words (phrases) or entities with limited descriptive content, or a text fragment composed of multiple entities in parallel, or a text fragment that may contain event information.

[0003] Most of the traditional classification strategies in existing research use the exhaustive method to determine candidate arguments, resulting in a huge number of candidate arguments, thus affecting the argument extraction effect; the machine reading comprehension strategy and the text generation strategy do not explicitly capture the associated semantics between roles; while the existing span selection strategy model does not fully exploit and utilize the roles and their associated semantics contained in event patterns and event instances. Summary of the Invention

[0004] In view of the above situation, the main object of the present invention is to propose a document-level event argument extraction method based on a role-focus heterogeneous graph to solve the above technical problems.

[0005] The present invention proposes a document-level event argument extraction method based on a role-focus heterogeneous graph, and the method includes the following steps:

[0006] Step 1: Associate event types with role nodes to obtain the associated semantics of the associated roles in the event pattern;

[0007] Step 2: Associate the trigger word and the central sentence in the event instance to obtain the associated semantics of the trigger word and the central sentence in the event instance, and associate the associated semantics of the associated roles in the event pattern with the associated semantics of the trigger word and the central sentence in the event instance to construct a role-focus heterogeneous graph;

[0008] Step 3: Input the role-focus heterogeneous graph into a graph attention network for learning to obtain the node representation after learning by the graph attention network;

[0009] Step 4: Use a pre-trained language model to capture document words for the node representation after learning by the graph attention network, and perform weighted fusion with the node representation after learning by the graph attention network to obtain a fused semantic representation;

[0010] Step 5: Construct an event argument extraction model based on the role focus heterogeneous graph, and use the event argument extraction model to perform cross-role argument extraction in combination with the fused semantic representation to obtain the probability of the start position of the role span in the document and the probability of the end position of the role span in the document;

[0011] Construct a loss function of the model based on the probability of the start position of the role span in the document and the probability of the end position of the role span in the document;

[0012] Optimize the event argument extraction model based on the loss function of the model to obtain an optimized event argument extraction model;

[0013] Step 6: Use the optimized event argument extraction model to obtain the event argument result.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0015] 1. The present invention designs a role semantic capture strategy for the role focus heterogeneous graph that reflects the semantic of the event structure. Capture the associated semantics of the role with the event type, trigger word, and central sentence from two levels of the event pattern and event instance, enrich the role semantics around the event structure, and improve the effect of span selection based on the role;

[0016] 2. The present invention develops a document-level event argument extraction model based on the role focus heterogeneous graph. By using a graph neural network, obtain the role representation updated through the role focus heterogeneous graph, and adopt a role-based span selection strategy to obtain the argument span acting as the role.

[0017] The additional aspects and advantages of the present invention will be partially given in the following description, partially become obvious from the following description, or be understood through the embodiments of the present invention. Description of the Drawings

[0018] Figure 1 It is a step flow chart of the document-level event argument extraction method based on the role focus heterogeneous graph proposed by the present invention.

[0019] Figure 2 It is a schematic diagram of the model framework of the document-level event argument extraction method based on the role focus heterogeneous graph proposed by the present invention. Detailed Embodiments

[0020] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals are the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation of the present invention.

[0021] These and other aspects of the embodiments of the present invention will be apparent from the following description and the accompanying drawings. In these descriptions and drawings, specific embodiments of the embodiments of the present invention are specifically disclosed as some ways of implementing the principles of the embodiments of the present invention, but it should be understood that the scope of the embodiments of the present invention is not limited thereto.

[0022] Please refer to Figure 1 , an embodiment of the present invention proposes a method for document-level event argument extraction based on a role-focus heterogeneous graph, and the method includes the following steps:

[0023] Step 1: Associate the event type with the role nodes to obtain the associated semantics of the associated roles in the event pattern.

[0024] In the above Step 1, associating the event type with the role nodes to obtain the associated semantics of the associated roles in the event pattern specifically includes the following steps:

[0025] Based on the event pattern, represent the event type and the role nodes in sets respectively, and obtain the set of event types and the role set respectively;

[0026] Establish an edge between the first event type and the second event type to obtain the associated semantics between the first event type and the second event type;

[0027] Establish an edge between the first event type and the first role node to obtain the associated semantics between the first event type and the first role node;

[0028] Establish an edge between the first role node and the second role node to obtain the associated semantics between the first role node and the second role node.

[0029] In the step of representing the event type and the role nodes in sets respectively to obtain the set of event types and the role set respectively, the following relational expressions exist for the corresponding process:

[0030] ;

[0031] Among them, represents the set of event types, represents the first event type, represents the node number of the event type, represents the number of event types included in the document, represents the th event type includes the role set, represents the first role node, represents the th event type includes the A role node, Indicates the number of roles included in the event type;

[0032] In the step of establishing an edge connection between the first event type and the second event type to obtain the associated semantics between the first event type and the second event type, the relational expressions existing in the corresponding process are as follows:

[0033] ;

[0034] Among them, Indicates the adjacency matrix corresponding to the role focus heterogeneous graph, Indicates the row or column of the corresponding node in the adjacency matrix, Indicates the th event type The set of roles included, Indicates the th event type The set of roles included;

[0035] In the step of establishing an edge connection between the first event type and the first role node to obtain the associated semantics between the first event type and the first role node, the relational expressions existing in the corresponding process are as follows:

[0036] ;

[0037] Among them, Indicates the th role node;

[0038] In the step of establishing an edge connection between the first role node and the second role node to obtain the associated semantics between the first role node and the second role node, the relational expressions existing in the corresponding process are as follows:

[0039] ;

[0040] Among them, Indicates the th role node.

[0041] Furthermore, establish an event type - event type edge connection, and connect the event type nodes that contain the same role name. Different event types may share roles with the same name, and through this type of edge, an association is established between different event types;

[0042] Establish an event type - role edge connection, and connect the event type node with each role node it contains. Different event types have different event templates and involve different role sets, and through this type of edge, the semantic information of the event type can be transmitted to the corresponding roles.

[0043] Establish role-role connections, where all role nodes under the same event type are connected pairwise. Different roles under the same event type are often used to describe different aspects of the same type of event. Through this type of edge, the associated semantics between event roles can be captured.

[0044] Step 2: Associate the trigger words and central sentences in the event instance to obtain the associated semantics of the trigger words and central sentences in the event instance. Then, associate the associated semantics of the associated roles in the event pattern with the associated semantics of the trigger words and central sentences in the event instance to construct a role-focus heterogeneous graph.

[0045] Please refer to Figure 2 , in Step 2, associate the trigger words and central sentences in the event instance to obtain the associated semantics of the trigger words and central sentences in the event instance. Then, associate the associated semantics of the associated roles in the event pattern with the associated semantics of the trigger words and central sentences in the event instance to construct a role-focus heterogeneous graph, which specifically includes the following steps:

[0046] Based on the event instance, represent the event trigger words and event central sentences included in the corpus respectively to obtain a set of trigger word nodes and a set of central sentence nodes;

[0047] Establish connections between the event trigger words and event central sentences to obtain the associated semantics between the event trigger words and event central sentences;

[0048] Establish connections between the event trigger words and the first event type to obtain the associated semantics between the first event trigger word and the event type;

[0049] Establish connections between the event trigger words and the first role node to obtain the associated semantics between the event trigger words and the first role node;

[0050] Establish connections between the event central sentences and the first event type to obtain the associated semantics between the event central sentences and the first event type;

[0051] Establish connections between the event central sentences and the first role node to obtain the associated semantics between the event central sentences and the first role node;

[0052] Based on the associative semantics between the first event type and the second event type, the associative semantics between the first event type and the first role node, the associative semantics between the first role node and the second role node, the associative semantics between the event trigger word and the event central sentence, the associative semantics between the first event trigger word and the event type, the associative semantics between the event trigger word and the first role node, the associative semantics between the event central sentence and the first event type, and the associative semantics between the event central sentence and the first role node, associate the event pattern with the event instance to form a heterogeneous graph of role focus.

[0053] In the step of respectively representing the event trigger words and the event central sentences included in the corpus to obtain the set of trigger word nodes and the set of central sentence nodes, the relational expressions existing in the corresponding process are as follows:

[0054] ;

[0055] Among them, represents the set of event trigger words, represents the first event trigger word, represents the th event trigger word, represents the number of trigger word nodes, represents the set of central sentence nodes, represents the th central sentence node, represents the number of central sentences, represents the first word of the central sentence node, represents the th word of the central sentence node, represents the number of words;

[0056] In the step of establishing an edge connection between the event trigger word and the event central sentence to obtain the associative semantics between the event trigger word and the event central sentence, the relational expressions existing in the corresponding process are as follows:

[0057] ;

[0058] In the step of establishing an edge connection between the event trigger word and the first event type to obtain the associative semantics between the first event trigger word and the event type, the relational expressions existing in the corresponding process are as follows:

[0059] ;

[0060] Among them, represents the event type corresponding to the trigger word node;

[0061] In the step of establishing an edge connection between the event trigger word and the first role node to obtain the associated semantics between the event trigger word and the first role node, the relational expressions existing in the corresponding process are as follows:

[0062] ;

[0063] In the step of establishing an edge connection between the event central sentence and the first event type to obtain the associated semantics between the event central sentence and the first event type, the relational expressions existing in the corresponding process are as follows:

[0064] ;

[0065] Among them, represents the event type of the event included in the central sentence node;

[0066] In the step of establishing an edge connection between the event central sentence and the first role node to obtain the associated semantics between the event central sentence and the first role node, the relational expressions existing in the corresponding process are as follows:

[0067] .

[0068] Furthermore, first establish the trigger word - central sentence edge connection, and connect the trigger word node of each event in the corpus with its corresponding central sentence node. Both the trigger word and the central sentence contain semantic information of the event instance to a certain extent. Through this type of edge connection, the key role of the trigger word in the central sentence can be enhanced;

[0069] Then, establish the pattern - instance association between the event pattern level and the event instance level to form a role - focus heterogeneous graph, and realize a deeper understanding of the semantic information of the role around the event structure in combination with the event instance. The specific steps are as follows:

[0070] Establish the trigger word - event type edge connection, and connect the trigger word node of each event in the corpus with the event type node of the event it triggers. The same type of events is triggered by multiple trigger words of the same type, and different types of events are usually triggered by different types of trigger words. Therefore, this type of edge can transfer the semantic information of the trigger word node to the event type node, thereby indirectly affecting the semantics of the role node;

[0071] Establish the trigger word - role edge connection, and connect each trigger word node in the corpus with all the role nodes of its corresponding event type. Through this type of edge, the trigger word semantics can be directly associated with the role, directly affecting the semantics of the role;

[0072] Establish the central sentence - event type edge connection, connecting each central sentence node in the corpus with the event type nodes it contains. Generally, the central sentence contains the core / subject information of the event, which can be used as event context to supplement the instance information of the event type and indirectly enrich the semantics of the role;

[0073] Establish the central sentence - role edge connection, connecting each central sentence node in the corpus with the role nodes corresponding to the events. Through this type of edge, the event context information can be associated with the role, directly enriching the semantics of the role corresponding to the event from the event instance.

[0074] Step 3: Input the role focus heterogeneous graph into the graph attention network for learning to obtain the node representation after being learned by the graph attention network.

[0075] In the said Step 3, when inputting the role focus heterogeneous graph into the graph attention network for learning to obtain the node representation after being learned by the graph attention network, the relational expressions existing in the corresponding process are as follows:

[0076] ;

[0077] Among them, represents the node representation after being learned by the graph attention network learning, represents the initial representation of the nodes of the graph, represents the adjacency matrix corresponding to the graph, represents the th role

[0078] representation after being learned by the heterogeneous graph network.

[0079] Step 4: Use the pre - trained language model to capture the document words for the node representation after being learned by the graph attention network, and perform weighted fusion with the node representation after being learned by the graph attention network to obtain the fused semantic representation.

[0080] ;

[0081] Among them, represents the set of document word representations, represents the representation of the first word, represents the th word represents the th word The representation indicating the first type of role The representation indicating the th type of role The representation indicating the encoding and decoding operations The representation indicating only the decoding operation The representation indicating the prompt The representation indicating the encoder output result;

[0082] In the step of performing weighted fusion on the node representations after learning by the graph attention network to obtain the fused semantic representation, the relational expressions existing in the corresponding process are as follows:

[0083] ;

[0084] Among them, The representation indicating the th type of role after fusion.

[0085] It should be noted that the role representations obtained by learning based on the role focus heterogeneous graph network are fused with the role representations in the Prompt obtained based on the pre-trained language model to further enrich the semantics of the roles.

[0086] Step 5: Construct an event argument extraction model based on the role focus heterogeneous graph, and use the event argument extraction model to perform cross-role argument extraction in combination with the fused semantic representation to obtain the probability of the start position of the role span in the document and the probability of the end position of the role span in the document;

[0087] Construct the loss function of the model based on the probability of the start position of the role span in the document and the probability of the end position of the role span in the document;

[0088] Optimize the event argument extraction model based on the loss function of the model to obtain the optimized event argument extraction model.

[0089] In the said Step 5, constructing an event argument extraction model based on the role focus heterogeneous graph, and using the event argument extraction model to perform cross-role argument extraction in combination with the fused semantic representation to obtain the probability of the start position of the role span in the document and the probability of the end position of the role span in the document specifically includes the following steps:

[0090] Perform span representation on the fused semantic representation to obtain the start representation of the role span and the end representation of the role span;

[0091] Use the event argument extraction model to predict the positions of the start representation of the role and the end representation of the role, so as to obtain the distribution of the start position of the role span on the document and the distribution of the end position of the role span on the document;

[0092] Perform probability calculations on the distribution of the start position of the role span on the document and the distribution of the end position of the role span on the document respectively, so as to obtain the probability of the start position of the role span on the document and the probability of the end position of the role span on the document.

[0093] In the step of performing span representation on the fused semantic representation to obtain the start representation of the role span and the end representation of the role span, the relational expressions existing in the corresponding process are as follows:

[0094] ;

[0095] Among them, represents the start representation of the role span, represents the end representation of the role span, represents the start vector, represents the end vector;

[0096] In the step of using the event argument extraction model to predict the positions of the start representation of the role span and the end representation of the role span, so as to obtain the distribution of the start position of the role span on the document and the distribution of the end position of the role span on the document, the relational expressions existing in the corresponding process are as follows:

[0097] ;

[0098] Among them, represents the role the distribution of the start of the span on the document, represents the role the distribution of the end of the span on the document;

[0099] In the step of performing probability calculations on the distribution of the start position of the role span on the document and the distribution of the end position of the role span on the document respectively, so as to obtain the probability of the start position of the role span on the document and the probability of the end position of the role span on the document, the relational expressions existing in the corresponding process are as follows:

[0100] ;

[0101] Among them, represents the distribution of the start position of the role span on the document, represents the distribution of the end position of the role span on the document, represents being normalized;

[0102] In the step of constructing the loss function of the model based on the probability of the start position of the role's span on the document and the probability of the end position of the role's span on the document, the relational expressions existing in the corresponding process are as follows:

[0103] ;

[0104] Among them, represents the model loss function, represents the number of documents included in the corpus, represents the documents included in the corpus, represents the logarithmic function, represents the number of documents, represents the number of roles involved in the event.

[0105] Step 6, Use the optimized event argument extraction model to obtain the event argument result.

[0106] In the said Step 6, the argument recognition F 1 value and the argument classification F 1 value are used as evaluation indicators. If an event argument span (i.e., the offset of the start and end positions in the document) and the event type match the span and event type mentioned by any argument, then the event argument is correctly recognized; if the role of the event argument is also correct, then the event argument is correctly classified. The relational expressions existing in the corresponding process are as follows:

[0107] ;

[0108] Among them, represents value, represents the precision rate, represents the recall rate, represents the number of true results predicted as true, represents the number of false results predicted as false, represents the number of true results predicted as false.

[0109] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following technologies well known in the art can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0110] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "examples", "specific examples", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0111] The above-described embodiments merely represent several implementation manners of the present invention. The descriptions are relatively specific and detailed, but should not be construed as limiting the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims

1. A document-level event argument extraction method based on role-focus heterogeneous graph, characterized in that: The method comprises the following steps: Step 1: associate the event type and the role node to obtain the association semantics of the associated role in the event pattern; Step 2: Associating the trigger words and the central sentence in the event instance, obtaining the associated semantics of the trigger words and the central sentence in the event instance, associating the associated semantics of the associated roles in the event pattern and the associated semantics of the trigger words and the central sentence in the event instance, and constructing a role focus heterogeneous graph; Step 3: Input the role focus heterogeneous graph into the graph attention network for learning, and obtain the node representation after learning by the graph attention network; Step 4: Use the pre-trained language model to capture document terms from the node representation learned by the graph attention network, and perform weighted fusion with the node representation learned by the graph attention network to obtain the fused semantic representation; Step 5: Construct an event argument extraction model based on the role focus heterogeneous graph, and use the event argument extraction model combined with the fused semantic representation to perform cross-role argument extraction to obtain the probability of the role's span start position on the document and the probability of the role's span end position on the document; The loss function of the model is constructed based on the probability of the character's span starting position on the document and the probability of the character's span ending position on the document; Optimize the event argument extraction model based on the loss function of the model to obtain the optimized event argument extraction model; Step 6: Use the optimized event argument extraction model to obtain the event argument results.

2. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 1 is characterized in that: In step 1, the event type and the role node are associated to obtain the associated semantics of the associated role in the event pattern, which specifically includes the following steps: Based on the event pattern, the event type and role nodes are represented by sets, and the event type set and role set are obtained respectively; Establish an edge between the first event type and the second event type to obtain the association semantics between the first event type and the second event type; Establish an edge between the first event type and the first role node to obtain the association semantics between the first event type and the first role node; An edge is established between the first role node and the second role node to obtain the association semantics between the first role node and the second role node.

3. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 2 is characterized in that: In the steps of respectively expressing the event type and the role node as a set and obtaining the set of event types and the set of roles, the corresponding processes have the following relational expressions: ； in, Represents a set of event types. Indicates the first event type, The node number indicating the event type. Indicates the number of event types contained in the document. Indicates Event Type A collection of characters included, Represents the first role node, Indicates The event type contains role nodes, Indicates the number of roles included in the event type; In the step of establishing an edge between the first event type and the second event type to obtain the association semantics between the first event type and the second event type, the relationship between the corresponding process is as follows: ； in, represents the adjacency matrix corresponding to the role-focus heterogeneous graph, represents the row or column of the corresponding node in the adjacency matrix, Indicates Event Type The character collection includes: Indicates Event Type Included character collection; In the step of establishing an edge between the first event type and the first role node to obtain the association semantics between the first event type and the first role node, the relationship between the corresponding process is as follows: ； in, Indicates role nodes; In the step of establishing an edge between the first role node and the second role node to obtain the association semantics between the first role node and the second role node, the relationship between the corresponding process is as follows: ； in, Indicates A role node.

4. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 3 is characterized in that: In step 2, the trigger word and the central sentence in the event instance are associated, the associated semantics of the trigger word and the central sentence in the event instance are obtained, the associated semantics of the associated role in the event pattern and the associated semantics of the trigger word and the central sentence in the event instance are associated, and a role focus heterogeneous graph is constructed, which specifically includes the following steps: Based on the event instance, the event trigger words and event central sentences contained in the corpus are represented respectively to obtain the trigger word node set and the central sentence node set; Establish an edge between the event trigger word and the event center sentence to obtain the associated semantics between the event trigger word and the event center sentence; Establish an edge between the event trigger word and the first event type to obtain the associated semantics between the first event trigger word and the event type; Establish an edge between the event trigger word and the first role node to obtain the associated semantics between the event trigger word and the first role node; Establish an edge between the event center sentence and the first event type to obtain the associated semantics between the event center sentence and the first event type; Establish an edge between the event center sentence and the first role node to obtain the associated semantics between the event center sentence and the first role node; According to the association semantics between the first event type and the second event type, the association semantics between the first event type and the first role node, the association semantics between the first role node and the second role node, the association semantics between the event trigger word and the event central sentence, the association semantics between the first event trigger word and the thing type, the association semantics between the event trigger word and the first role node, the association semantics between the event central sentence and the first thing type, and the association semantics between the event central sentence and the first role node, the event pattern is associated with the event instance to form a role focus heterogeneous graph.

5. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 4 is characterized in that: In the step of respectively representing the event trigger words and the event central sentences contained in the corpus to obtain the trigger word node set and the central sentence node set, the corresponding process has the following relational expression: ； in, Represents a set of event trigger words, Indicates the first event trigger word, Indicates Event trigger words, Indicates the number of trigger word nodes, represents the set of central sentence nodes, Indicates Central sentence nodes, Indicates the number of central sentences, Indicates the first word of the central sentence node, Indicates the central sentence node words, Indicates the number of words; In the step of establishing an edge between the event trigger word and the event center sentence to obtain the associated semantics between the event trigger word and the event center sentence, the relationship between the corresponding process is as follows: ； In the step of establishing an edge between the event trigger word and the first event type to obtain the associated semantics between the first event trigger word and the event type, the relationship between the corresponding process is as follows: ； in, Indicates the event type corresponding to the trigger word node; In the step of establishing an edge between the event trigger word and the first role node to obtain the associated semantics between the event trigger word and the first role node, the relationship between the corresponding process is as follows: ； In the step of establishing an edge between the event center sentence and the first event type to obtain the associated semantics between the event center sentence and the first event type, the relationship between the corresponding process is as follows: ； in, Indicates the event type of the event contained in the central sentence node; In the step of establishing an edge between the event center sentence and the first role node to obtain the associated semantics between the event center sentence and the first role node, the relationship between the corresponding process is as follows: 。 6. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 5 is characterized in that: In step 3, the role focus heterogeneous graph is input into the graph attention network for learning, and the node representation after learning by the graph attention network is obtained. The relationship between the corresponding process is as follows: ； in, Represents the graph attention network The learned node representation is express The initial representation of the nodes of the graph, express The adjacency matrix corresponding to the graph is, Indicates Type of role Representations after learning with heterogeneous graph networks.

7. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 6 is characterized in that: In step 4, the pre-trained language model is used to capture document terms on the node representation learned by the graph attention network. The relationship between the corresponding process is as follows: ； in, Represents a document word representation set, Indicates the expression of the first word, Indicates Words The expression, Indicates the document Words The expression, Indicates the first type of role. Indicates The representation of a role, Represents encoding and decoding operations, Indicates that only decoding operations are performed. Indicates the prompt. express Encoder The output result of In the step of weighted fusion of node representations learned by the graph attention network to obtain the fused semantic representation, the corresponding process has the following relationship: ； in, Indicates Type of role The fused representation of .

8. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 7 is characterized in that: In step 5, an event argument extraction model is constructed based on the role focus heterogeneous graph, and the event argument extraction model is combined with the fused semantic representation to perform cross-role argument extraction, so as to obtain the probability of the role's span start position on the document and the probability of the role's span end position on the document, which specifically includes the following steps: Perform span representation on the fused semantic representation to obtain a span start representation and a span end representation of the character; The event argument extraction model is used to predict the positions of the span start representation and the span end representation of the role, so as to obtain the distribution of the span start position and the span end position of the role in the document; Probability calculations are performed on the distribution of the span start position of the character on the document and the distribution of the span end position of the character on the document to obtain the probability of the span start position of the character on the document and the probability of the span end position of the character on the document.

9. The document-level event argument extraction method based on role-focus heterogeneous graph according to claim 8 is characterized in that: In the step of performing span representation on the fused semantic representation to obtain the span start representation of the character and the span end representation of the character, the relationship between the corresponding processes is as follows: ； in, The span that indicates the role begins to indicate, The span that indicates the role ends. represents the starting vector, Indicates the end vector; In the step of using the event argument extraction model to predict the position of the span start representation of the role and the span end representation of the role to obtain the distribution of the span start position of the role on the document and the distribution of the span end position of the role on the document, the corresponding process has the following relationship: ； in, Representing roles the distribution of span starts over the documents, Representing roles Distribution of span endings over the document; In the step of respectively calculating the probability of the distribution of the span start position of the character on the document and the distribution of the span end position of the character on the document to obtain the probability of the span start position of the character on the document and the probability of the span end position of the character on the document, the relationship between the corresponding processes is as follows: ； in, represents the distribution of span start positions of roles on the document, represents the distribution of span end positions of roles on the document, It means that it has been normalized; In the step of constructing the loss function of the model based on the probability of the character's span start position on the document and the probability of the character's span end position on the document, the corresponding process has the following relationship: ； in, represents the model loss function, Indicates the number of documents contained in the corpus, Indicates the documents contained in the corpus. represents the logarithmic function, Indicates the number of documents. Indicates the number of roles involved in the event.

Images