Databases excel at managing transactions such as adding, updating, and deleting records, and are adept at retrieving specific records using primary keys, etc. On the other hand, search engines capable of full-text search use inverted indexes, allowing them to rapidly execute diverse and flexible searches across all fields (equivalent to columns in a database table) of all documents (records) as follows:
- Exact matching of keywords (words or phrases), fuzzy search, prefix matching, and suffix matching.
- Search using complex query expressions with Boolean operators like AND/OR/NOT and parentheses.
- Processing of synonyms.
- Filter query.
- Sorting search results by similarity.
KandaSearch also supports semantic search, which allows searching based on the 'meaning' of the documents being searched, in addition to providing flexible and fast keyword search functionality as described above. The search targets documents (not limited to text, but also including images, videos, audio data, etc.), and it can search based on the 'meaning' they possess.
However, unlike databases, search engines are not proficient in transaction management provided by databases. Therefore, it is necessary to build applications that leverage the characteristics of both.
