Word pair index structure for information retrieval using Terrier3.5

With the evolution of internet, users are looking for documents that are accessible on demand. The growth in volume of stored information affects the performance of retrieval systems. In any information retrieval system, two main factors to be considered are document representation termed as indexing and retrieval. Most of the queries to an information retrieval system are casual, ordinary, bag of words that consists of two or three terms. In this paper, we implemented an index structure for fast phrase querying namely Word pair index using Terrier 3.5, a well-known open source for information retrieval. In our experiments, we used FIRE 2011 English data set. As we increased the size of corpus, time to create index structure has also increased linearly. When the word pair index structure is used for retrieval, the time to retrieve relevant results from the corpus reduced by 94.73%, when compared to traditional inverted index structure.