Large-scale related entity network-based related entity recommendation method

Abstract

The invention relates to a large-scale related entity network-based related entity recommendation method. The method comprises the following steps of: S1, related entity network construction: extracting large-scale related entities from mass entity description pages in the internet, and constructing a large-scale related entity network by utilizing a description/described relationship between title entities and body entities; S2, entity relevancy calculation: calculating entity relevancy by adoption of an entity concept and co-adjacent entity-based semantic similarity and a graph structure-based link analysis technology on the basis of the large-scale related entity network, and taking the entity relevancy as a feature for related entity sorting; and S3, related entity recommendation: recommending related entities for a query entity on the basis of the above two steps. According to the method, millions of high-quality entities and ten million-level entity relationships can be rapidly and effectively mined on the basis of mass Web texts, so as to construct the cross-field, large-scale and high-quality related entity network; and the method is capable of remarkably enhancing the entity recommendation correctness and novelty.

Description

technical field [0001] The invention relates to a method for recommending related entities based on a large-scale related entity network, and belongs to the technical field of information retrieval. Background technique [0002] Since the emergence of Internet search engines in the 1990s, Internet search has become an indispensable part of people's daily life. The current search engine is mainly based on keyword matching retrieval mode, resulting in user queries are usually short in length, noisy, and ambiguous. Query recommendation aims to recommend to users queries related to the query they entered. This technology can not only help users locate the search intent, but also provide users with different queries related to the original query, guide or predict the user's next query, and is an important method to improve user search experience and improve search engine service quality. [0003] As an important part of Web query, entity query is very common in vertical search ...

AI Technical Summary

Problems solved by technology

Due to the short length of entity queries and possible ambiguity, it is very difficult to accurately capture the user's query intent in this case

If only the correlation between the recommended content and the original entity query is considered, it is likely to cause redundant recommendations

However, most of the existing research works use query logs or query related documents as data sources, and use correlation calculation methods based on query terms, query co-clicks, or query co-occurrences to recommend related queries, and pay more attention to the relevance of recommended content. , but there is less research on guiding and inspiring users' click interests

Images