Evaluation the accuracy and recall in general search engines, based on the system relevance and search logic

Document Type : Research Paper




Purpose: using search engines is a way to retrieve information. The aim of this study was to evaluate and compare the recall and accuracy of general search engines, based on the system relevance and search logic.
Methodology: This comparative study was done using three key words” scientometry”, “scientific production” and “citation analysis” which based on searched keywords in IranDoc are the most frequent during 1389-1393 in the field of information sciences. These keywords were searched in Google- Yahoo- Bing- Ask- AOI using Boolean search logic operation (AND), fuzzy logic (the short way) and vector model in English. According to search logic first 10 results per search engine were saved then the recall and accuracy were calculated.
Findings and Conclusion: In all three logic search Google had the highest level of accuracy and recall (accuracy: 0.833 ± 0. 20 and recall: 0.125 ±0.03) and Bing had the lowest levels (Accuracy: 0.277± 0.12 and recall: 0.41 ± 0.02), ASK was at the second level in all three methods and AOL and Bing had the weakest function in vector model and truncation respectively. Google with Boolean logic had the most and with truncation had the least relevant documents. ASK in the three search logic had a same level of relevancy, Yahoo with Boolean logic search had the lowest and with vector model had the highest number of relevant documents. Bing had the lowest relevant documents with truncation and eventually AOL was at the lowest operating level with vector model.