Learning-To-Rank Low-Level Feature Library

A library for the collection of common low-level features used in learning-to-rank algorithms. These features are calculated using well-defined algorithms.

If you feel there are additional features that might be added to this library, create a new issue or submit a pull request.

High level features created using machine learning based methods can be found or added to my highletorfeats library.

Features

The following categories of low-level features are collected using this library:

• TF-IDF
• Okapi BM25
• LMIR (TODO)

Getting started

Adding to your project

The recommended way to add this library to you project is by including the following to your CMakeLists.txt:

cmake_minimum_required(VERSION 3.13)
project(myProject)

include_directories("path/to/lowletorfeats/include")
add_subdirectory("path/to/lowletorfeats")

add_executable(myProject myProject_SOURCES)
# or `add_library(myProject myProject_SOURCES)`

target_link_libraries(myProject lowletorfeats)

Usage

The library has can be included in your source files with the following:

#include <lowletorfeats/FeatureCollector.hpp>

An example case can be found in tests/src/main.cpp.

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

• Patrick Cox - paddy74

License

This project is licensed under the MIT License - see the LICENSE.md file for details.