Effective Lightweight Learning-to-Rank Method Using Unified Term Impacts

Abstract

In this study, we propose and evaluate a novel learning-to-rank (L2R) approach that produces results on par with those of the state-of-the-art L2R methods while being computationally effective. We start by presenting a modified gradient boosted regression tree algorithm to generate unified term impact (UTI) values at indexing time. Each unified term impact replaces several features with a single value in the document index, thereby reducing the effort to compute the document scores at query processing time because the system fetches and processes fewer values. The adoption of UTI values produces competitive ranking results. However, the lack of features available only at query time might lead to accuracy loss. To solve this problem, we propose a hybrid model that uses UTI values with query-dependent features. We demonstrate that our hybrid methods can deliver high-quality results on par with those of the existing state-of-the-art neural ranking models. Our methods can also reduce the computational costs for processing queries, serving as an interesting alternative for L2R practical applications. Our best hybrid, HLamb-daMART, achieves an NDCG@10 value of 0.495 using only 36 features at query processing time when applied to the MQ2007 collection, while the best baseline achieves 0.490 using a larger set of features at query processing time. The use of our hybrid framework reduces the time to run LambdaMART to about 35% of the time to run it without using our proposals. In summary, we present a competitive and lightweight alternative L2R approach to be adopted in search systems.

Relevance ranking for proximity full-text search based on additional indexes with multi-component keys

“…In modern search systems, a two-level relevance calculation process is often used [26,27]. Two ranking mechanisms are applied one after another.…”

“…Назовем данный метод R IntervalOpt . В современных системах часто применяется двухуровневый расчет релевантности [26,27]. Последовательно применяются два механизма ранжирования.…”

“…Каждому документу или результату поиска соответствует вектор чисел. Второй механизм можно реализовать на основе машинного обучения, например искусственных нейронных сетей [28], генетических алгоритмов [29], градиентного бустинга [27]. При построении функции релевантности используется тестовая коллекция документов с рассчитанными векторами признаков и рангами документов для какого-то набора поисковых запросов.…”

Jusbrasil e os Desafios Tecnológicos para Facilitar e Aprimorar o Acesso à Justiça

“…Além disso, sistemas de busca disponíveis no mercado levam muito tempo para incorporar melhorias propostas na literatura, em parte porque algumas delas podem não beneficiar a todos os tipos de empresas que usam tais sistemas ou ainda podem até prejudicar parte de seus usuários. Exemplos de melhorias que podem impactar muito sistemas de busca como o da Jusbrasil, mesmo que não impactem tanto a outros tipos de sistemas de busca incluem novos métodos de poda e processamento de consultas [Rossi et al 2013, Daoud et al 2017, Petri et al 2019] e novas técnicas para o desenvolvimento de sistemas de busca eficientes que usem aprendizagem de máquina [Lucchese et al 2018, Silva et al 2020].…”

A Knowledge-Fusion Ranking System with an Attention Network for Making Assignment Recommendations

“…FAIR-PG-Rank recommends items via a policy-gradient approach which could satisfy fairness of exposure constraints with respect to items [ 31 ]. A similar idea in the article is generating unified term impact (UTI) during the indexing time and combining into a hybrid model to improve the accuracy [ 32 ]. Since the relationship between study performance and exam results is much complicated, article [ 33 ] finds the correspondence of input values and predicts targets which is not a one-to-one relation, treats the classification task as a fuzzy geometrical problem, and proposes a fuzzy similarity approach to solve the problem [ 34 , 35 ].…”