The third episode of the second season of Lucidworks’ of our explainer video series Lucid Thoughts is ready and waiting. This week we’re looking at neural information retrieval search or neural IR search for short.
Neural information retrieval search uses deep learning algorithms and machine learning techniques and discern query intent and context. Traditional keyword search is still useful but this advanced approach is the next step towards improving precision and relevancy, espcially for applications with big data sets like collections of research journals or legal docuents.
Low-level neural IR methods use user behavior to calculate which word or sentence is the most important in a collection of raw text documents. Synonym detection is also how the machine can learn to associate particular words or when they occur in a common fashion
One popular algorithm is the Word2Vec ranking model. Word2Vec turns words and phrases into vectors on a graph and then uses trigonometric calculations to find the distance between specific words and phrases. The closer two phrases are in the graph – and in the documnt – the more likely they align to the user’s query.
You can binge-watch all of season two of Lucid Thoughts or jump on back to watch season one.
Subscribe to the Lucid Thoughts channel and be sure to leave your questions and comments on each video.
Leave a Reply