Text Recommendation Using Self-Attention and Graph Convolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing knowledge graph convolutional network (KGCN) models for text recommendation suffer from low accuracy due to random initialization of initial entity vectors, ignoring the fusion of text data, leading to suboptimal recommendation results.
Innovation Solution
A text recommendation method utilizing a self-attention model and a trained graph convolutional network model, incorporating a pre-built text knowledge graph and historical text retrieval information, to determine interaction probabilities and recommend targeted texts based on semantic information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If random initialization is used for entity, user, and relationship vectors in KGCN models, then the model structure is simple and easy to implement, but the recommendation accuracy is low due to neglecting text data fusion
Solution Approach 1:
The patent applies preliminary action by pre-building a text knowledge graph before recommendation. The knowledge graph is constructed in advance by extracting entities and relationships from text data, storing them in a structured format. This preliminary construction allows the recommendation system to leverage rich text semantic information without adding complexity during the actual recommendation process, thus improving accuracy while maintaining model simplicity.
Solution Approach 2:
The patent uses an intermediary approach by introducing a text knowledge graph as a mediator between text data and the recommendation model. The knowledge graph serves as an intermediate representation that captures semantic relationships from text, which are then fed into the KGCN model. This intermediary structure enables the model to utilize text information effectively without directly processing raw text, thereby improving recommendation accuracy while keeping the core model structure relatively simple.
2Measurement precision
If text embedding vectors are integrated into the knowledge graph convolutional network, then the fusion of text data improves recommendation accuracy, but the processing time and computational complexity increase
Solution Approach 1:
The patent applies preliminary action by pre-computing text embedding vectors and constructing the knowledge graph in advance. The text data is processed, embedded, and organized into the knowledge graph structure before the recommendation process begins. This preliminary preparation reduces the computational burden during actual recommendation queries, as the system only needs to retrieve and process pre-processed information rather than performing all computations in real-time, thus improving accuracy while mitigating processing time costs.
Data Source
AI summary
Provided in the present disclosure are a text recommendation method and apparatus, a model training method and apparatus, and a readable storage medium. The text recommendation method includes: acquiring text retrieval information from a user; when it is determined that there is historical text retrieval information for the user, determining text information of each text in a text set retrieved by using the text retrieval information; performing embedded representation on the text information of each text based on a self-attention model, and determining a text embedding vector of each text; inputting the text embedding vector of each text into a trained graph convolutional network model, to obtain the probability of interaction between the user and each text in the text set; and screening out, from the text set, target text which meets a preset interaction probability, and recommending the target text to the user.