Retrieval Effectiveness of News Search Engines: A Theoretical Framework

News search has now become an important internet activity as users are switching from hard copies to online news reading. Many modern news search engines like: Google News or Bing News are available for this purpose. We propose a theoretical framework for evaluating the retrieval effectiveness of news search systems. The framework exploits supervised machine learning approach for evaluating therefore we performed retrieval effectiveness tests on a small data set consisting relevancy featuresTfidf and Latent Semantic Indexing (LSI) as well as freshness feature-publication time, extracted from 1120 query-document pairs collected from search results of Google News, to evaluate the performance of various machine learned learning to rank algorithms on NDCG and ERR metric at different cut-offs. The motive behind this work is to conduct large-scale retrieval effectiveness studies for news search engines. General Terms Feature engineering, machine learning, retrieval effectiveness tests