Language problem in retrieving images on web search engines

Information sources are increasing rapidly and exponentially on the Web. Many kinds of information sources from all over the world can be accessed easily through search engines. Although Web is successful to overcome the geographical barriers in access to information, language is still a problem to overcome. In other words the main issue is not retrieving the documents in various languages, but to be able to utilize from them. Many search engines provide “translation support” so that their users are able to find and use text based sources in foreign languages. But in case of searching for images, they don’t provide “translation support” and give the results depending on search keywords. Therefore the general judgment which depends on the current belief that Web will be a solution for digital divide, is not completely realistic yet. The validity of this idea may be considered as true for image sources also, because images are more effective than text based documents in conveying their content. The aim of this study is to show that language is still a barrier in retrieving and using images. In this study the language barrier/problem for queries in different languages on different search engines is tested. According to this, information retrieval performance of Google, Yahoo! and Bing are evaluated in terms of precision ratio, relative recall ratio and language independence ratio with first 10 and 20 cut-off points of total results. According to our findings, search engines have language barrier/problem in retrieving images. Solution which is presented, to overcome this problem may be providing multilanguage information retrieval by using cross language information retrieval method and visual features