Uporaba metod obdelave naravnega jezika za uvrščanje shem programskih vmesnikov v enoten podatkovni model

Conradi, Matic (author), Žitnik, Slavko (mentor)

Cilj te raziskave je razviti sistem, ki je zmožen klasificirati sheme objektov JSON, ki izvirajo iz različnih sistemov za digitalizacijo procesov, ter najti ujemanja s predstavitvami domenskih entitet platforme DevRev. Glavni izziv, s katerim se soočamo, leži v raznolikosti struktur teh objektov, kar zahteva uporabo algoritmov za obdelavo naravnega jezika.

• Language: Slovenian
• Keywords: obdelava naravnega jezika, uvrščanje shem, Word2Vec, BERT, GPT, veliki jezikovni modeli
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2023
• PID: 20.500.12556/RUL-148956
• COBISS.SI-ID: 164523267
• Publication date in RUL: 01.09.2023

Secondary language

• Language: English
• Title: Use of natural language processing methods for classification of application programming interface schemas into a unified data model

Abstract

The goal of the thesis is to develop a system for classification JSON object schemas originating from various process digitalization systems, and ultimately matching them with domain entities from the DevRev platform. The primary challenge lies in the varying structures of these objects, which necessitates the use of NLP algorithms to capture and compare them.

• Keywords: natural language processing, schema classification, Word2Vec, BERT, GPT, large language models