This is an “unusual” research aspect of Information Retrieval (IR). By trying to compare and analyze different IR models in a formal way, Axiomatic Framework can show some interesting and even astonishing results of IR models. For instance, it can show that IR models should satisfy certain number of constraints. If a model cannot satisfy some of them, we can expect its performance being worse. This is the type of comparison without any experiments at all, though the claims are indeed justified by empirical studies.

Materials:

• Axiomatic Analysis and Optimization of Information Retrieval Models by ChengXiang Zhai at ICTIR11 [Slides]
• Yuanhua Lv, ChengXiang Zhai. Lower-Bounding Term Frequency Normalization. Proceedings of the 20th ACM International Conference on Information and Knowledge Management  (CIKM’11), 2011. [PDF]
• Hui Fang, Tao Tao, and Chengxiang Zhai. 2011. Diagnostic Evaluation of Information Retrieval Models. ACM Transactions on Information Systems (TOIS) 29, 2, Article 7 (April 2011), 42 pages. [PDF]
• Hui Fang, ChengXiang Zhai, Semantic Term Matching in Axiomatic Approaches to Information Retrieval. Proceedings of the 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval ( SIGIR’06 ), pages 115-122. [PDF]
• Hui Fang, ChengXiang Zhai, An Exploration of Axiomatic Approach to Information Retrieval. Proceedings of the 28th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval ( SIGIR’05 ), 480-487, 2005. [PDF]
• Hui Fang, Tao Tao, ChengXiang Zhai, A formal study of information retrieval heuristics. Proceedings of the 27th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval ( SIGIR’04), pages 49-56, 2004. [PDF]
• Hui Fang‘s PhD dissertation. [PDF]