Statistical Machine Translation Is Done By Conjunction

1612 Words7 Pages
Problem Statement
Matching corresponding terms in queries and documents in the corpus is done by conjunction. Due to this, recall is low in case of long queries. We can redevelop such queries to improve the probability of matching corresponding terms in ueries and relevant documents. This process is called query expansion. In this research paper,various models that take as input user query records and output replacements for query terms from terms in the revelent documents, are analyzed. The best outlook to employ while training the model to learn from the training data is also described.
Solution Approach
Statistical Machine Translation is a machine transaltion method in which translations from foreign to English language are made based on the statistical models whose parameters are derived from the study of tet corpus from both the languages. When the corpus consists of a single language its called monolingual statistical machine translation. In this research paper, we also compare different models employed for query expansion. The method we employ is:
*Snippets from the relevant document set i.e., short text pieces are taken from the highest ranked results to user queries.
*they are paired with the user queries, which is referenced user query records maintained from user sessions.
*We train and apply the monolingual Statistical machine Translation Model on these query-snippet pairs using a machine learning algorithm.
*We get as output expansion

More about Statistical Machine Translation Is Done By Conjunction

Get Access