SAJARI
Self Learning
User feedback drives algorithm optimisation. Results adjust in a local sense (users current search) as well as in a global sense (algorithm evolves and improve results for all).
User feedback drives algorithm optimisation. Results adjust in a local sense (users current search) as well as in a global sense (algorithm evolves and improve results for all).
Ever wanted to upload a document and find similar ones in seconds? Now you can. The biggest query we've supported so far is ~50 pages of text.
A type of Bayesian learning that detects relationships between input text and core concepts. This means the actual input text does not need to appear in the results. For instance a search for "tax" may detect results with "accounting" or "audit".
Every object can optionally have a latitude and longitude. Search requests can geofence and look for results in specified regions, or gracefully degrade result importance as distance increases from a geopoint X-Y.
Filter by date, title, location or any other meta field you choose to add. Any numeric meta fields can use operators such as <, >, >=, <=, =. Other uses include userID, segmentation information, categories, weight, length, salary, etc.
Supports uploading of entities (locations, names, skills, job titles, industries, etc) and automatically detects these in search and recommendation functions. Can be used to parse input text to find entities, or boost results if certain entities are detected.
Any free text can be uploaded to Sajari. Text will be indexed and stored automatically. Meta information can also be added to any record. There is also no restriction on meta keys, add as many as you like.
Fully API driven. Supports JSON response format over HTTP. Almost all engine functions can be achieved via our API.