Research has provided indications as to whether search engine results are biased. The study was carried out by using news search results of the Google and Bing search engines, and 57 controversial subjects were evaluated in terms of two types of bias, namely “perspective” and “ideology”. The article resulting from the research was published in Information Retrieval, one of the most prestigious journals in this field.
The research carried out by Yücel Saygın, member of Sabancı University Faculty of Engineering and Natural Sciences, and Gizem Gezici, PhD Student at the Computer Science and Engineering Program, in collaboration with Aldo Lipani and Emine Yılmaz from University College London (UCL), on the impartiality of search engines was published in Information Retrieval as one of the leading studies in this field.Gizem Gezici, Yücel Saygın
ARE SEARCH ENGINES UNBIASED?
It is search engines that decide which results are listed in response to a query. Among the most frequently used tools to access information nowadays, search engines are expected to be unbiased on the basis of the sources of information they present. However, search engine results do not cover all the perspectives about a search query subject. Search engine results can be biased in favor of a certain perspective since search results are listed according to level of relevance. Therefore, it is important to evaluate search engine results from the perspective of impartiality.
The focus of the research of Professor Yücel Saygın, member of Sabancı University Faculty of Engineering and Natural Sciences, and Gizem Gezici, PhD Student at the Computer Science and Engineering Program, concerned the evaluation of whether search engines brought the most relevant results in response to user queries. The 57 controversial issues used in the search queries were taken from ProCon.org, an online resource for research on controversial issues.
SEARCH ENGINES WERE OBSERVED TO COME UP WITH BIASED RESULTS
The study was conducted by using the news search fields of the Google and Bing search engines, and results were evaluated in terms of two types of bias, namely “perspective” and “ideology”. The researchers used these two types of bias to see whether the search engines brought biased results separately. In addition, the two search engines were compared in terms of level of bias. The conclusion of the study was that the search engines did not include any bias about perspective, however both search engines came up with ideologically biased results. Furthermore, it was observed that the more successful search engine came up with more biased results. The findings of the research established that it was necessary to include metrics for bias in success criteria for search engines.