Query understanding is the process of inferring the user’s intent by extracting semantic information from the query’s keywords. It is a core component of any search engine and can have a significant impact on the search results.
Search queries have certain unique characteristics compared to documents, which makes them difficult to handle: queries do not always observe the syntax of a written language and contain limited context.
In my internship project, I developed a machine learning model for segmenting search queries into meaningful tokens (phrases) and disambiguating the phrases with concepts and entities. The challenge was to develop a model that can deal with a very large vocabulary (many millions) of concepts and entities. The model improved the understanding of intents and products in the Google Shopping search queries.