Case affirmation method and system for performing causal inference based on correlation graph information
A correlation diagram and cause of action technology, applied in the field of cause-of-action identification methods and systems, can solve problems such as violation of causality and characteristic errors of learning, and achieve the effects of improving accuracy, fast training speed, and fewer model parameters.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0060] Such as figure 2 As shown in , a method for determining the cause of action based on correlation graph information for causal inference, including:
[0061] S1: Obtain the factual description of the case;
[0062] S2: Extract key information according to the factual description of the case; including using the KeyBERT algorithm to obtain the keywords of the case, and cluster the keywords of the case.
[0063] S3: Use the GFCI algorithm to construct a causal graph and sample a causal subgraph;
[0064] S4: Estimate the causal strength of the edges of the causal subgraph, and combine the BIC algorithm to obtain the causal strength of the causal graph, and obtain the judgment result;
[0065] S5: Use the causal strength of the causal graph to construct an auxiliary loss, improve the performance of the BiLSTM-Att model, and obtain a trained BiLSTM-Att model;
[0066] S6: For the identified case, obtain the factual description of the case, input it into the trained BiLST...
Embodiment 2
[0069] According to the method for identifying the cause of action based on the correlation diagram information described in Embodiment 1, the difference is that:
[0070] The specific implementation process of S1 is as follows:
[0071] Obtain the factual description of the case, including various facts and specific circumstances of the case, including the subject of the case, generally an individual or an organization. There are also the behavior and various circumstances of the subject of the case, as well as the harmful results caused, that is, the object and specific object of the behavioral harm and the fact of its severity, the causal relationship between the behavior and the result. Beyond that there is the motive and purpose of the case.
Embodiment 3
[0073] According to the method for identifying the cause of action based on the correlation diagram information described in Embodiment 1, the difference is that:
[0074] In S2, for highly structured document types such as company documents, the KeyBERT algorithm is used to extract key information in the description, including using BERT to extract document embeddings to obtain document-level vector representations, and then extract word vectors for N-garm words . Then, use cosine similarity to find the words most similar to the document. This algorithm calculates the importance of each word to the case. At the same time, in order to distinguish the different cases of the case, the most important p words for the case, that is, the p words with the highest cosine similarity, are selected, and they are aggregated into q types of keywords with similar characteristics, including using the K-means clustering algorithm Cluster these words. The specific implementation process of ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More - R&D
- Intellectual Property
- Life Sciences
- Materials
- Tech Scout
- Unparalleled Data Quality
- Higher Quality Content
- 60% Fewer Hallucinations
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.
© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com