Sparque Desk Documentation

Getting Started
Information that will help you to get started with SPARQUE.
Datasets
Information on creating knowledge graphs from data sources, defining mappings with SPARQUE Language, and loading triples into Sparque's graph database.
Strategies
Information on managing strategies using the Strategy Editor, connecting and configuring building blocks, setting API outputs, adding parameters, and creating stacked strategies for complex search tasks.
- Text Transformations for Search Term Matching
- Building Blocks
APIs
Information on managing APIs in SPARQUE Desk to connect strategies to your application.
Evaluation
Information on assessing strategy performance in SPARQUE Desk using testsets and evaluations.
Users
Information on managing user permissions and roles.
Settings
Information on managing your account settings.
Query API
Information on using the Query API. Learn about endpoints, faceted search, response types, and request options.
Miscellaneous
Information on Tuple-notation, encoding, error messages and short-hand notation.
Shopware Extension
The SPARQUE Shopware extension allows you to seamlessly integrate our powerful tools into your Shopware store, providing advanced data analytics and insights.
SPARQUE API Wrapper
Information on the SPARQUE API Wrapper, which streamlines interactions between clients, internal users, and SPARQUE Desk.
SPARQUE Hub
Information on SPARQUE Hub, the front-end component designed to simplify the interactions of clients and internal users with SPARQUE Desk.
Project Tasks
Information that will help you perform SPARQUE project tasks.
Glossary
Discover important terms and concepts used throughout SPARQUE Desk.

Home
Strategies
Information on managing strategies using the Strategy Editor, connecting and configuring building blocks, setting API outputs, adding parameters, and creating stacked strategies for complex search tasks.
Building Blocks
Rank
IDF [Obj,Obj]

Getting Started
Information that will help you to get started with SPARQUE.
Datasets
Information on creating knowledge graphs from data sources, defining mappings with SPARQUE Language, and loading triples into Sparque's graph database.
Strategies
Information on managing strategies using the Strategy Editor, connecting and configuring building blocks, setting API outputs, adding parameters, and creating stacked strategies for complex search tasks.
- Text Transformations for Search Term Matching
- Building Blocks
APIs
Information on managing APIs in SPARQUE Desk to connect strategies to your application.
Evaluation
Information on assessing strategy performance in SPARQUE Desk using testsets and evaluations.
Users
Information on managing user permissions and roles.
Settings
Information on managing your account settings.
Query API
Information on using the Query API. Learn about endpoints, faceted search, response types, and request options.
Miscellaneous
Information on Tuple-notation, encoding, error messages and short-hand notation.
Shopware Extension
The SPARQUE Shopware extension allows you to seamlessly integrate our powerful tools into your Shopware store, providing advanced data analytics and insights.
SPARQUE API Wrapper
Information on the SPARQUE API Wrapper, which streamlines interactions between clients, internal users, and SPARQUE Desk.
SPARQUE Hub
Information on SPARQUE Hub, the front-end component designed to simplify the interactions of clients and internal users with SPARQUE Desk.
Project Tasks
Information that will help you perform SPARQUE project tasks.
Glossary
Discover important terms and concepts used throughout SPARQUE Desk.

IDF [Obj,Obj]

Description

Computes IDF of T in DT object pairs

Input

DT [OBJ,OBJ]: a list of object-object pairs.

Output

DTidf [OBJ,OBJ]: IDF-weighted pairs
Tidf [OBJ]: IDF-weighted objects

Parameters

IDF: the IDF weighting scheme variant. P(t|d) = t_nd / d_N, n_D = number of documents, t_nd = number of documents with term t.
- BASIC: non-logarithmic IDF (1 - P(t|d) = 1 - (t_nd / d_N))
- IDFB: non-smoothed IDF (-log( P(t|d) = log ( d_N / t_nd ))
- IDFP: smoothed IDF (log( (d_N - t_nd + 0.5) / (t_nd + 0.5) ))
- IDFPA: alpha-smoothed IDF (alpha * IDFP) Output scores can be aggregated and/or normalized.