Study On Information Retrieval Algorithm Guided With Query Conceptual Graph

As the Internet and information technology is developing fast,the demand of fast and high quality information retrieval is becaming more and more critical.At the beginning,we discussed the current status of traditional and semantic search techniques.To provide a better info retrival service,we need to grasp users' demand better.In this paper,we discussed the characteristics and differences between traditional search engine and the questions from Q&A websites.The analysis showed that users have huge demand for query service in question form.In this paper,We first gave the representation form,mathematical definition and calculation rules of conceptual graph.then we discussed the computing of similarity of two conceptual graphs.At last we paid attention on how to retrieve and mark conceptual graph up from texts.After that,we fixed our mind on exploring methods on info retrieval with given conceptual graphs.We combine RDD computing model with our conceptual graph model and comes up with a method to convert a conceptual graph into RDD computing graph model.We first find the center node of the conceptual graph,from it we makes a spanning tree with BFS.Then we trim extra relation nodes and turn concept nodes along with the corresponding relation nodes into Transformation node in pairs.At last we add Action nodes to gather the results.With this computing stucture we can get the final rank score for each document that flows over the graph.A prototype system is designed and implemented for this algorithm.We performed benchmark and measured serveral searching scores compared to extended boolean model.The results showed that our method is feasible and efficient.The model discussed in this paper explored the method to make use of conceptual graphs to perform information retrieval actions and is proved to be feasible and efficient.This model may provide some inspirations for further research.