Information retrieval (IR) seeks to support users in finding information relevant to their information needs. One obstacle for many IR algorithms to achieve better results in many IR tasks is that there is insufficient information available to enable relevant content to be identified. For example, users typically enter very short queries, in text-based image retrieval where textual annotations often describe the content of the images inadequately, or there is insufficient user log data for personalization of the search process. This thesis explores the problem of inadequate data in IR tasks. We propose methods for Enriching Information Retrieval (ENIR) which address various challenges relating to insufficient data in IR. Applying standard methods to address these problems can face unexpected challenges. For example, standard query expansion methods assume that the target collection contains sufficient data to be able to identify relevant terms to add to the original query to improve retrieval effectiveness. In the case of short documents, this assumption is not valid. One strategy to address this problem is document side expansion which has been largely overlooked in the past research. Similarly, topic modeling in personalized search often lacks the knowledge required to form adequate models leading to mismatch problems when trying to apply these models improve search. This thesis focuses on methods of ENIR for tasks affected by problems of insufficient data. To achieve ENIR, our overall solution is to include external resources for ENIR. This research focuses on developing methods for two typical ENIR tasks: text-based image retrieval and personalized web data search. In this research, the main relevant areas within existing IR research are relevance feedback and personalized modeling. ENIR is shown to be effective to augment existing knowledge in these classical areas. The areas of relevance feedback and personalized modeling are strongly correlated since user modeling and document modeling in personalized retrieval enrich the data from both sides of the query and document, which is similar to query and document expansion in relevance feedback. Enriching IR is the key challenge in these areas for IR. By addressing these two research areas, this thesis provides a prototype for an external resource based search solution. The experimental results show external resources can play a key role in enriching IR.