Knowledge Graph Query Expansion for Search Recall Improvement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information retrieval systems struggle to effectively capture relevant information due to the lack of recognition of synonymous or related keywords, leading to suboptimal search recall.
Innovation Solution
A knowledge graph is constructed for an enterprise to identify synonyms and related words by structuring information into interconnected nodes and edges, using graph random walk and graph convolutional network algorithms to enhance keyword embeddings, enabling the search engine to retrieve additional related keywords.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional keyword matching is used for information retrieval, then the system is simple and fast, but search recall is poor because synonymous or related keywords are not recognized
Solution Approach 1:
A knowledge graph is introduced as an intermediary between the user query and the document retrieval system. The knowledge graph contains pre-computed embeddings for entities, attributes, and values extracted from enterprise data, serving as a mediator to expand and refine search queries by identifying synonymous and related terms before retrieving documents.
Solution Approach 2:
The system performs preliminary actions by pre-extracting entities, attributes, and values from enterprise data and pre-computing their embeddings before actual search operations. This advance preparation creates the knowledge graph structure that enables faster and more accurate query expansion during actual search operations.
2Reliability
If a knowledge graph with graph random walk and graph convolutional network algorithms is constructed to identify synonyms and related words, then search recall is significantly improved, but the system complexity and computational resources increase
Solution Approach 1:
Traditional mechanical keyword matching is replaced with graph-based computational methods. Graph random walk algorithms simulate random traversals through the knowledge graph to discover related terms, while graph convolutional networks use neural network operations on graph structures to learn and represent semantic relationships, substituting simple string matching with sophisticated computational models.
Solution Approach 2:
The system changes the parameter representation from simple keyword strings to multi-dimensional embedding vectors. By transforming discrete keywords into continuous vector representations that capture semantic meaning, the system enables similarity-based matching and expands the search space to include conceptually related terms beyond exact keyword matches.
Data Source
AI summary
Systems and methods of knowledge graph construction are provided. A communication platform accesses communication data on the communication platform associated with an enterprise. The communication platform extracts a plurality of keywords from the communication data. The communication platform creates a knowledge graph comprising a plurality of nodes. The plurality of nodes comprises a plurality of keyword nodes. The communication platform identifies, using a graph random walk algorithm, one or more keyword nodes in the knowledge graph related to a first keyword node to link the first keyword node with the one or more keyword nodes. The communication platform determines a first keyword node embedding for the first keyword node by aggregating information associated with at least the first keyword node and the one or more keyword nodes using a graph convolutional network algorithm. The communication platform attaches the first keyword node embedding to the first keyword node in the knowledge graph.