A Keyword Search ksanew Method Combining Semantic Nodes and Edge Weights

Abstract

The invention relates to a keyword search KSANEW algorithm combining semantic nodes and edge weights. Including: the data storage stage, which dynamically updates the data such as semantic classes, entities and attributes in the knowledge base with the arrival of knowledge fragments; the keyword query stage, which takes into account the model layer of the knowledge graph compared to the data layer data The amount is small, and a query seed model is proposed, which maps query keywords to the schema layer, and then generates candidates through two types of expansion methods, namely node-based expansion in the direction of large weights and edge-based expansion in the direction of large weights The seed model, and then use the scoring function to score and sort the candidate seed set, and finally use the high-scoring candidate seeds as query seeds, and use the query seeds as a guide to perform distributed searches on the data layer to obtain query results.

Description

technical field [0001] The invention belongs to the technical field of mass data storage and retrieval under a knowledge graph, and in particular relates to a keyword search KSANEW method combining semantic class nodes and edge weights. Background technique [0002] At present, the calculation methods of keyword weights are mainly divided into two categories: the calculation methods based on the characteristics of the keywords themselves and the calculation methods based on the text structure characteristics. The calculation methods based on the characteristics of keywords themselves mainly include: TF-IDF method, CHI method, IG method and semantic-based method. The biggest disadvantage of the TF-IDF weight calculation method is that the keyword is regarded as an independent individual, ignoring the additional weight generated by the combination of keywords; the semantic-based calculation method although the weight calculation is based on the essential characteristics of the...

AI Technical Summary

Problems solved by technology

The biggest shortcoming of the TF-IDF weight calculation method is that keywords are regarded as independent individuals, ignoring the additional weights generated by the combination of keywords; although the semantic-based calculation method starts from the essential characteristics of keywords to calculate the weight , but depends on the accuracy of the semantic dictionary, the ischemic nature of the semantic dictionary will restrict the accuracy of this calculation

The calculation method based on the structural characteristics of the text defines the keyword weights based on the structural characteristics of the text. This method is more effective for texts with a relatively regular structure, but relatively ineffective for texts with chaotic structures.

[0003] Since the arrival of knowledge fragments will make the knowledge base dynamically change, the timeliness of the knowledge base will also become one of the key points of consideration. However, the existing keyword weight calculation methods do not fully consider the timeliness characteristics. Therefore, the present invention is based on the background of knowledge graphs. Next, a keyword search algorithm combining semantic nodes and edge weights is proposed. This algorithm combines the time-sensitive characteristics with the weight calculation formulas of semantic nodes and edges, and builds a query seed model with time-sensitive characteristics on the knowledge map model layer. , and then use the query seed as a guide to perform distributed keyword retrieval on the data layer to obtain query results

Images