SIRA: arquitetura para recuperação de informação sem servidor

Abstract

The growing need for efficient and scalable information retrieval systems has driven the search for new approaches and architectures. This work proposes the Serverless Information Retrieval Architecture (SIRA) for information retrieval on the serverless computing platform, focusing on the stages of indexing, searching, and classification. The architecture is divided into two steps: document indexing, which includes processing, metric extraction, and the creation of inverted indexes; and document retrieval, which involves searching for relevant documents, classifying the documents through metrics, and presenting the results to the user. The performance of the SIRA architecture is compared to Elastic Search (ELS) on two test collections, Cranfield and TREC-COVID. In addition to having the advantage of consuming resources only on demand, the SIRA architecture demonstrated similar performance to ELS, indicating its effectiveness in retrieving relevant documents and its potential as a solution in the field of information retrieval.