How Taruqa evaluates queries - Fluid Topics

AFS Taruqa

Category
Technical Notes
Audience
public

In Taruqa, query evaluation is the probability that the query will be generated by the documents of the corpus. If metadata is used, it also fits into the probability model. In the Taruqa model, a feature is any element that can be used in the evaluation process (that is part of the equation).

Ultimately, all criteria can be described as a probability score:

Probability tree

Taruqa offers a consistent query language to express how the score of each document is computed. The functions to select and score documents are not hard-coded or hidden in the indexing or document evaluation process. Taruqa dynamically transforms the user's query - expressed in a high-level and understandable query language - into a search equation.

Query compilation

It is possible to define this search equation by configuring search behavior so that it integrates default preferences (latest product version, reference guides first, etc.), as well as preferences based on information about the user gathered from their past searches, profile, etc.

Personalized search

With Taruqa, queries take any number of textual and non-textual parameters into account to provide personalized search results.