A framework for information quality and coverageassessment for type 2 diabetes websites

Ölçer, Didem
Health information seekers often use search engines to access high-quality and up-to-date information. However, finding online high-quality health information is increasingly getting difficult due to the high volume of information generated by non-experts in the area. There are manual tools which help end-users in assessing the quality of websites. However, they are labour-intensive. This thesis aims to propose a framework that automatically evaluates the content coverage and quality of health websites according to evidence-based medicine. The thesis has two main contributions. The first one is a method which utilizes quality indicators derived from professional health literacy guidelines to measure information quality. The second contribution includes a method which uses both textual and content-based features with Okapi BM25 and MeSH term expansion to assess information coverage and information quality. Content-based features were acquired using American Diabetes Association’s (ADA) guideline, which is an evidence-based practice guideline in diabetes. Specifically, sentences containing auxiliary verbs from ADA guideline were extracted and the weirdness coefficient of terms, 2-grams and 3-grams generated from these sentences were calculated using iWeb corpus. The results showed that the use of both textual and content-based features is effective in classification of high and low-quality websites. In addition, the features derived from professional health guidelines lead to a significant positive impact in classification results.
Citation Formats
D. Ölçer, “A framework for information quality and coverageassessment for type 2 diabetes websites,” Ph.D. - Doctoral Program, Middle East Technical University, 2020.