Probabilistic Indexing Search Extensions

Except for a brief section at the end of Chapter 5, so far in this book PrIx has been studied assuming queries are just single, individual words. However, in the practical PrIx applications, users expect to be allowed to formulate usual forms of free-text searching queries. Typically, these forms encompass multiple words, presented as word sequences and/or logically combined using Boolean operators. Moreover, flexible word spelling is also typically expected, mainly to cope with the user incognizance of exactly how a given word could have been written in the manuscripts – but also to deal with the system (or human) uncertainty as to which pen strokes were actually used to depict some characters or words. In this chapter we consider in detail these and other related topics, from a mostly practical point of view which puts effectiveness and efficiency issues before perfect mathematical formulation or exactness of the equations.