Comparing Ranking Learning Algorithms for Information Retrieval Systems.

The growth of machine learning applications in various fields has enabled the advancement of Information Retrieval systems. As a result of this evolution, it has become possible to solve the well-known document classification problem. In the beginning, document positions within a result were given from a score, where each document receives an assigned value based on the terms used in the input query. The use of machine learning in this field is known as Learning to Rank, which allows the classification of documents to better meet user search requirements, taking into account aspects such as document preference, importance, and relevance. This paper presents a comparison of different algorithms for ranking documents using machine learning. It is observed that RankSVM presents relatively satisfactory results in smaller datasets, while algorithms that use Gradient Boosting obtain better results for larger datasets.