An Event Template Construction Method Based on Entity Connectivity Graph

A construction method and technology of connected graphs, applied in the field of information processing, can solve problems such as difficulty in extracting key information of texts, difficult quantitative calculation of text similarity, inaccurate recognition results and specific information extraction

Active Publication Date: 2021-07-16
HARBIN INST OF TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the inaccuracy of identifying results and extracting specific information in the application of artificial intelligence such as "retrieval", "question and answer" and "dialogue" for specific events in the prior art; it is difficult to quantitatively calculate the similarity of texts based on events And the problem that the key information of the text is difficult to extract, and a method of event template construction based on entity connectivity graph is proposed

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • An Event Template Construction Method Based on Entity Connectivity Graph
  • An Event Template Construction Method Based on Entity Connectivity Graph
  • An Event Template Construction Method Based on Entity Connectivity Graph

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0028] Specific implementation mode one: combine figure 1 To illustrate this embodiment, the specific process of a method for constructing an event template based on an entity connectivity graph in this embodiment is as follows:

[0029] The goal has two points: 1) Determine the relevance of the events stated between the input texts, and divide the input text set into multiple clusters accordingly, and the texts in the clusters all state the same event; 2) According to the text described in the clusters, Information, to obtain event templates. Event templates are divided into two types: coarse-grained and fine-grained. Coarse-grained event templates only need to identify the trigger word of the event, the people involved in the event, and time (that is, identify the event elements), and fine-grained The event template automatically extracts the event element according to the information described in the text in the cluster, and obtains the corresponding value of the event elem...

specific Embodiment approach 2

[0062] Specific embodiment 2: the difference between this embodiment and specific embodiment 1 is that in the step 4, word vectors are used to calculate S 1 The similarity between any two nodes, the similarity calculation formula is:

[0063]

[0064] In the formula, θ is the vector A i with B j angle, A i is a word vector (A 1 ,A 2 ,...,A n ) of the i-th number (any node of an article), B j is a word vector (B 1 ,B 2 ,...,B n ) of the jth number (any node of an article), n is a positive integer; (A 1 ,A 2 ,...,A n ) is a node.

[0065] Other steps and parameters are the same as those in Embodiment 1.

specific Embodiment approach 3

[0066] Embodiment 3: This embodiment differs from Embodiment 1 or Embodiment 2 in that: the range of the threshold in step 4 is 0.4-0.8.

[0067] Other steps and parameters are the same as those in Embodiment 1 or Embodiment 2.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

A method for constructing an event template based on an entity connection graph, and the invention relates to a method for constructing an event template. The purpose of the present invention is to solve the inaccuracy of identifying results and extracting specific information in artificial intelligence applications such as "retrieval" and "question answering" for specific events in the prior art; it is difficult to quantitatively calculate the similarity of texts based on events and difficult to Extraction problem. The process is: 1. Extract triples for each sentence; 2. Treat the three elements in the triples as three nodes; 3. Put all the nodes extracted from the text into the set S 1 Middle; 4. Calculating S using word vectors 1 The similarity between any two nodes and generate a connected graph accordingly; 5. Calculate the PageRank value of each node; 6. Determine the relationship measurement value between the events stated in the two texts; 7. Form multiple text clusters; 8. , Build coarse-grained event templates and fine-grained event templates. The invention is used in the fields of text key information extraction and correlation degree calculation.

Description

technical field [0001] The invention relates to a construction method of an event template and relates to the technical field of information processing. Background technique [0002] With the rapid development of Internet technology, the Internet is flooded with a large number of events, most of which exist in the form of unstructured text, such as "XX discovered XX", "XX acquired XX", etc. This unstructured Text information is difficult to be applied to artificial intelligence technologies such as "retrieval", "question answering", and "dialogue". Therefore, how to automatically obtain the relevance of the events stated in the text from the network text to divide the text into multiple event clusters and then construct an event template is very meaningful. [0003] Although many evaluation organizations (such as ACE) have defined the types of events and the structured representation of events, for example, ACE defines 8 major types of events and 33 small types of events, b...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F16/35G06F16/36
CPCG06F16/353G06F16/36
Inventor 秦兵刘铭刘挺刘一仝李嘉伟
Owner HARBIN INST OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products