Abstract

When “Search” is mentioned, the first thing come to mind is “Web Search” and users will think in Google, Bing, or Yahoo. This is true since it is the most common search task by normal users. However, when talking about search with its scientific meaning “Information Retrieval”, specialists can list tens of tasks other than web search. Among these tasks are “Printed Documents Search” and “Patent Search”, which are highlighted in this talk. These two search tasks differs from web search in nature, where they have different objectives, type of users, and retrieval challenges. Two different approaches for searching printed documents are presented. The first works on overcomming the errors of the recognized text, where different methods for achieving this are discussed. The second approach works on searching this printed document in its image form without doing any recognition. For the patent search task, an introduction to its nature and challenges is discussed, and novel approaches for the evaluation of this task and for patent cross-language search are presented.