Research On A Method Of Mining User’s Search Intent Based On Knowledge Graph

The principle of computer to understand the user’s search intent is very complex. Because, for the same query, different people’s intent have huge difference; for the same search intent, different people may use completely different query. In order to achieve a particular query intent to understand, can analyze query log and establish interest model, but it does not help recognition the new query term, because the user’s intention is not fixed theme, the needs of intent is changing. Even if a user enters a query which constitute by the same word, it is not represent of the user want to know about the same topics, resources, or network services.Category-based search intent mine is the mainstream of mining search intent. First, establish the classification system of intent. Second classify the query text into categories. The current mainstream category system of search intent has been built to be very perfect, but because the user’s query usually is very short, with a certain ambiguity, and its classification feature is limited. In which case, the text classification is difficult. Now, there is not an effective method of extracting classification feature to solve this situation.This paper proposes a method of mining search intent bases on knowledge graph, make use of knowledge graph inter-word association to adjust the weight of query keyword can improve the accuracy of classification, and improve excavation effect of search intent.After the experiment, compared with several classical method of weight adjustment, our method can effectively enhance the accuracy of classification. Our method can help mining search intent very well. After follow-up study, I believe, using this method for interactive information retrieval and man-machine conversation will also achieve a good effect.