Semantic Information Browse and Search Application - bounswe/bounswe2024group5 GitHub Wiki

Semantic Search

Semantic Search is a type of search engine that interprets the meaning of words. The key here is to match the search results to the contextual meaning behind a search query, instead of the literal matches of the words in the query. This way, the search returns more relevant results semantically.

Such search engine approach aims to improve the search quality by interpreting natural language in context. This semantic interpretation can be achieved by utilizing machine learning technologies and NLP.

In semantic search, context refers to any information, from the searcher's location to her search history, to the words in the search query. Semantic search utilizes context clues to determine the meaning of a search qery across a dataset of millions of examples.

Vector Search

Vector search uses machine learning to capture the meaning and context of unstructured data and transform it into a numeric representation. Vector search utilizes k-nearest nearest neighbor (kNN) algorithm to find similar data. Then, it generates results and ranks them based on conceptual relevance. Semantic search often uses vector search to provide more relevant results more efficiently.

Why Choose Semantic Search?

From the users' perspective, a search engine that uses the previous data of the user provides semantically relevant information. Thus, it facilitates more intuitive search experience. Such user specific search will provide individual results to the user, hence makes them feel 'heard'. Also, semantically relative search results have higher chance to meet customers' expectation of the search. Thus, the searching process will be viewed more efficient in the customer's eyes. Moreover, a semantic search algorithm continues to "learn" from the user data. Such continuously improving user satisfaction aspect is what makes semantic search preferable than regular search engines.

Some Semantic Search Examples

Summary Notes

  • Semantic search provides more relevant search results.
  • It utilizes NLP and ML technologies like kNN to find semantically related data.

Some Resources for Further Learning

Author

Ebrar