| Internet technology flourishes in recent years and the explosion of information put forward higher requirements to retrieve information and resources. Thanks for the high development of the network technology and the hard work of scientists, all respect of information retrieval technology and the performance of search engine has been greatly improved.Mongolian is the main national language and the official language in Inner Mongolia, which is valuable information and wealth. Along with large numbers of Mongolian website’s easily available, the Mongolian People urgently need to improve their search efficiency for the massive Mongolian information. The smart Mongolian people create a wide variety of coding to express their written words, such as Mengkeli, saiin, Mongolian international standard code and so on and all of them have theirs own unique feature. But the main coding on the web is mengkeli coding, which can not be directly applied to the domestic or foreign search platform. To solve this problem and improve the efficiency of Mongolian searching, we propose our new methods based on the previous research.We applied the new incremental approach of query expansion and proposed the extraction of extensional terms based on the binned abstract in order to further improve the efficiency of search. The basic idea of the new incremental approach is to reduce the time needed for the second round of retrieval by using the accumulator of the first retrieval.And then we can avoid to retrieve the same word appeared in both the first and extensional query twice and improve the efficiency of search program. This reduces the overall pseudo-relevance feedback retrieval time and improve the efficiency of the Mongolian retrieval. We also proposed a new idea to extract the extensional terms based on the binned abstract. Experimental result on Mongolian corpus shows the combination of those two methods can improve the efficiency of Mongolian search program. |
PDF Full Text Request