文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > IT/计算机 > 互联网 > An Empirical Evaluation on Semantic Search Performance of Keyword-Based and Semantic Search Engines
侵权投诉

An Empirical Evaluation on Semantic Search Performance of Keyword-Based and Semantic Search Engines

With increase in cut-off point value, Google’s precision ratios decreased for all cut-off points. Hakia’s precision ratio slightly increased at cut-off point 10, then decreased slightly at cut-off points 15 and 20. Furthermore, Google retrieved more relevant documents than Msn for all cut-off points with approximately 3% difference. Although Yahoo retrieved approximately 4% more relevant documents than Hakia at cut-off points 10, 15 and 20, it retrieved the highest number of relevant documents at all cut-off points and its best precision ratio was 48% at cut-off point 5. However, Msn retrieved the least number of relevant documents at all cut-off points. Generally, precision ratios of search engines decreased when the cut-off point values were increased.

An Empirical Evaluation on Semantic Search Performance of Keyword-Based and Semantic Search Engines

Figure 2. Mean normalized recall ratios of search

engines

Figure 2 shows mean normalized recall ratios of keyword-based search engines and those of the semantic search engine at various cut-off points. Yahoo’s normalized recall ratio was approximately the same as that of Msn at cut-off point 5. However, for cut-off points 10, 15 and 20, while the normalized recall ratio for either of these search engines increased, the other search engine’s normalized recall ratio decreased. Normalized recall ratios of Google and Msn were approximately the same at cut-off point 15 and the difference between Google, Yahoo and Msn was about 1% at cut-off point 20. Google had the highest performance at cut-off point 5 (88%) but when the cut-off point increased to 10 and 15, Google’s normalized recall ratio decreased. However, Google had the same normalized recall ratio at cut-off points 15 and 20. At all cut-off points, Google had the highest performance for displaying relevant documents retrieved in the top ranks of the retrieval output. Although Hakia’s normalized recall ratio increased gradually at all cut-off points, Hakia had the least performance for displaying relevant documents retrieved in the top ranks of the retrieval output.

4. Conclusion and Future Work

In this paper, an investigative evaluation on search performance of keyword-based and semantic search engines is detailed. It was found that Google, Yahoo and Msn retrieved at least one relevant document for all queries, whereas Hakia3 failed to retrieve any relevant document for one of the queries. Although Hakia4 retrieved more relevant documents compared to other search engines, Hakia3 retrieved least relevant documents.

In terms of overall performance, Hakia retrieved more relevant documents compared to Google and Msn at all cut-off points. However, Yahoo retrieved the highest number of relevant documents at all cut-off points with its best precision ratio being 48% at cut-off point 5. Google showed the highest performance for displaying relevant documents in the top ranks of the retrieval output at all cut-off points. Yahoo and Msn come next, while Hakia showed the least performance for displaying relevant documents retrieved in the top ranks of the retrieval output.

Generally, precision ratios of search engines decreased with increased cut-off point values. However, it was seen from the results that the performance of search engines, when displaying relevant documents in the top ranks, is better than their relevant document retrieval.

Finally, it was seen that semantic search performance of search engines was low regardless of the type of the search engine used. Therefore, search engines need to improve their systems, taking into consideration the importance of the role semantic search can play in helping users getting precise information from the web with minimal effort.

As a future work, the most frequently used queries and phrases could be run on search engines. Furthermore, the number of search engines, queries and phrases could be increased. In addition, elaborate statistical analysis could be provided.

5. References

[1] Google, doc.guandang.net (Accessed on 09 February 2009)

[2] Yahoo, doc.guandang.net (Accessed on 09 February 2009)

[3] Msn, doc.guandang.net (Accessed on 09 February 2009)

[4] Hitwise, “Hitwise US – Leading Search Engines”, October 2008. doc.guandang.net (Accessed on 12 November 2008)

上一页第4页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email