Causal Knowledge Graph Generation With NLP Attribute Labels
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques in knowledge mining from textual data fail to effectively incorporate attribute labels such as topic, sentiment, and temporal information, limiting the ability to perform targeted searches and visualize unidirectional branches in causal knowledge graphs, which hampers the exploration of entity trajectories and insights.
Innovation Solution
A method and system that determine causal chains of events using NLP, assign attribute labels (topic, sentiment, and temporal labels) to entities, and generate a knowledge graph by clustering nodes to interlink them through directional edges, enabling richer insights and visual exploration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If current knowledge mining techniques are used, then basic causal relationships can be extracted, but attribute labels (topic, sentiment, temporal) cannot be effectively incorporated
Solution Approach 1:
The patent segments the knowledge graph construction process into distinct modules: causal relationship extraction, attribute label assignment (topic, sentiment, temporal), and integrated graph generation. This segmentation allows each component to be optimized independently while maintaining overall system functionality, enabling effective incorporation of attribute labels without compromising causal relationship extraction.
Solution Approach 2:
The patent merges multiple functionality into a unified knowledge mining system that simultaneously extracts causal relationships and assigns attribute labels. By combining these previously separate tasks into an integrated framework, the system recovers lost attribute information while maintaining versatility in search capabilities across multiple dimensions.
2Adaptability or versatility
If comprehensive attribute labels are assigned to all entities, then search capability is enhanced, but the complexity of the knowledge graph increases
Solution Approach 1:
The patent applies local quality by assigning attribute labels selectively to entities based on their specific characteristics and relevance to causal relationships. Rather than uniformly labeling all entities, the system applies different types of labels (topic, sentiment, temporal) only where appropriate, enhancing search capability while avoiding unnecessary complexity in the overall graph structure.
Solution Approach 2:
The patent adds multiple dimensions of attribute labels (topic, sentiment, temporal) to the knowledge graph structure. This dimensional enrichment enables sophisticated multi-dimensional searches without fundamentally complicating the underlying graph architecture, as the attributes operate as additional layers of organization rather than structural modifications.
3Loss of information
If unidirectional branches are visualized in the knowledge graph, then entity trajectories can be explored, but the visualization complexity increases
Solution Approach 1:
The patent employs asymmetry in visualizing unidirectional branches by representing causal relationships with directed edges that clearly indicate the direction of influence from cause to effect. This asymmetric representation makes entity trajectories easily traceable without requiring complex bidirectional navigation, simplifying visualization while preserving trajectory information.
Solution Approach 2:
The patent performs preliminary organization of entities and their attribute labels before visualization. By pre-processing and structuring the data with assigned attributes and identified causal relationships, the system simplifies the subsequent visualization step, making it easier to render unidirectional branches and explore entity trajectories without dealing with raw unstructured data.
Data Source
AI summary
The present disclosure relates to a method for generating a knowledge graph. The method includes determining a causal chain of events indicating a cause-and-effect relationship among entities within the input data based on a causal expression. Further, the method includes assigning attribute labels such as a topic label, a sentiment label, and a temporal label to the entities using a Natural Language Processing (NLP) technique. Further, the method includes creating nodes indicating a collection of entities having the assigned attribute labels. Furthermore, the method includes generating a knowledge graph based on clustering the nodes. The knowledge graph indicates a visual depiction of the causal chain of events such that the nodes are interlinked through a directional edge representing the causal chain of events. In the method, the generated knowledge graph along with the assigned attribute labels is retrieved based on at least one of a user-query input or parameter filters.