The exemplary embodiments of the present invention provide a system, method and computer program products for a snippet based proximal search. A method comprises ranking the documents based on text that is similar to a text snippet. The ranking includes automatically generating proximity queries that include the text snippet, submitting the proximity queries, and collecting the document results. The method comprises selecting a plurality of highest ranked documents to form a subset of documents, extracting snippets from each document in the subset, and creating a vector space model for a set defined by a union of the extracted snippets and the text snippet. The method comprises ranking the extracted snippets according to their vector distances from the input text snippet, and ranking the documents within the subset of documents based on the ranking of the extracted snippets to determine a best matching document in view of the text snippet.