Language Models and Smoothing Methodsfor Information RetrievalVon der Fakultät für Ingenieurwissenschaften,Abteilung Informatik und angewandte Kognitionswissenschaft,der Universität Duisburg-Essenzur Erlangung des akademischen Gradeseines Doktors der Ingenieurwissenschaften (Dr.-Ing.)genehmigte DissertationvonNajeeb A. Abdulmutalib M.Scaus Ben Walid (Libya)Gutachter:Prof. Dr.-Ing. Norbert FuhrProf. Dr.-Ing. Gerhard WeikumTag der mündlichen Prüfung: 29. Oktober 2010AbstractDesigning an effective retrieval model that can rank documents accurately for a givenquery has been a central problem in information retrieval for several decades. Anoptimal retrieval model that is both effective and efficient and that can learn fromfeedback information over time is needed. Language models are new generation ofretrieval models and have been applied since the last ten years to solve many differentinformation retrieval problems. Compared with the traditional models such as thevector space model, they can be more easily adapted to model non traditional andcomplex retrieval problems and empirically they tend to achieve comparable or betterperformance than the traditional models. Developing new language models is currentlyan active research area in information retrieval.In the first stage of this thesis we present a new language model based on an oddsformula, which explicitly incorporates document length as a parameter.