[A Semantic Meta-Search Engine]

By

Table of Contents

CHAPTER 4 SOLUTION1

CHAPTER 5: IMPLEMENTATION, RESULT, FUTURE WORK10

CHAPTER 6 CONCLUSION14

END NOTES17

CHAPTER 4 SOLUTION

Most Web users use search engines to find the information they want from the Web. One common complaint about the current search engines is that they return too many useless results for users' queries. Both the search engines and the users contribute to this problem. On the one hand, current search engines make little effort to understand users' intentions and they retrieve documents that match query words literally and syntactically. On the other hand, Internet users tend to submit very short queries (average length is about 2.3 terms and 30% have a single term [8]). One way to tackle this problem is to group the search results for a query into multiple categories such that all results in the same category corresponds to the same meaning of the query. In this paper we propose a new technique to group the search result records (SRRs) returned from any search engine. Our focus will be on SRRs retrieved by single term queries. For queries with multiple terms, the specific meaning of each term is easier to determine because other terms in the same query can provide the context [9]. Our technique differs from existing techniques in the following aspects. First, we use a semantic electronic dictionary WordNet [5, 13] to provide the basic meanings of each query term. Second, we apply a merging algorithm to merge synsets that have very close meanings into a super-synset. Third, we employ a two-step process to categorize SRRs into super-synsets. Fourth, our method also deals with SRRs that do not correspond to any WordNet-provided synsets of the query terms by clustering them. For example, when a word is used as a name, like “Apple” and “Jaguar”, it does not have its traditional meanings.

With the vertiginous increase of volume information on the Web, the results pro-vided by traditional search engines in response to user queries do no longer satisfy the needs of specific communities of users. There is an increasing amount of an-swers that satisfies the terms contained in user queries. However, these answers are not precise enough for some users demanding a more refined list of results ac-cording to the semantics of their queries. This open problem has motivated a new era of search systems that have received the name of Semantic Search Engines. A Semantic Search Engine (SSE) can be understood as a semantic Web appli-cation that can answer questions based on the meaning of users query specifica-tion, resources in the repositories and in many cases it is based on predefined domain semantics or a knowledge model. SSE can return relevant results on your topics that do not necessarily mention the word you searched for explicitly. The goal of this work is to study and discuss various widespread research direc-tions in semantic search engines, as well as identifying common features and main approaches used in them. In this work we can get an overview of current ap-proaches to ...