Extracting Trends from Incomplete Ordinal Preferences Vahid M. Hashemi Abstract: In conventional preference aggregation, usually the result of aggregating multiple preferences is a single preference. Thus, the preference of the majority is seen as the aggregated preference of the whole society. But often, members of a society have many different tastes and these tastes are partly recognizable from their preferences. There are many applications in which having more than one preference as the aggregation of the input preferences is more useful (or necessary). Another aspect of such applications is that the agents may not have a preference ordering consisting of all candidates (this is not actually feasible when there is a very large number of candidates, e.g., all music or movies on a huge database), and they would only be required to submit partial preferences on the set of candidates. In this thesis we suggest a new approach in computational social choice to have more than one preference as the aggregated preference of the society. We present a model to extract these aggregated preferences (we call them trends) from incomplete ordinal preferences. Furthermore, we introduce a number of axiomatic properties to evaluate this new model and use them to investigate the properties of our proposed methods for this model.