Analiza podobnosti pesmi na podlagi besedil, ocen in meta podatkov

Babnik, Žiga (author), Šubelj, Lovro (mentor)

Robustnost in natančnost glasbenega priporočilnega sistema je povezana s kvaliteto in tipom podatkov, ki jih ta upošteva. Različne vrste podatkov se razlikujejo po zahtevnosti analize in pridobivanja. Podatke, ki jih je težje pridobiti in analizirati, želimo nadomestiti s podatki, ki nosijo enako informacijo in jih je lažje pridobiti in analizirati. Analiziramo podatkovne nabore besedil pesmi, ocen pesmi in glasbenih meta podatkov. Iz podatkovnih naborov zgradimo matrike podobnosti pesmi s pomočjo tekstovnega rudarjenja, analize omrežij in analize vektorjev. Matrike primerjamo z različnimi merami in rezultate primerjanj zapišemo v matrike podobnosti naborov. Pregled matrik podobnosti naborov omogoča odkrivanje skritih podobnosti med različnimi glasbenimi podatki. Izkaže se, da je podobnost med nabori omejena na tip podatkov in način analize podatkov.

• Language: Slovenian
• Keywords: glasba, podobnost glasbe, priporočilni sistem, tekstovno rudarjenje, analiza omrežij, analiza vektorjev, podobnost matrik, gručenje
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2019
• PID: 20.500.12556/RUL-110202
• COBISS.SI-ID: 1538346179
• Publication date in RUL: 12.09.2019

Secondary language

• Language: English
• Title: Song similarity analysis based on lyrics, ratings and meta data

Abstract

The accuracy and robustness of song recommendation systems depends on the quality and type of data given to the system. Different types of data vary in difficulty of extraction and analysis. We wish to replace data which is harder to extract and analyse, with data that carries the same information, yet is easier to extract and analyse. The extracted data sets include song lyrics, song popularity scores and song meta data. From the data sets we build song similarity matrices for each data set, using text mining, network analysis and vector analysis. Song similarity matrices are then compared using five different measures, and the results are stored in data set similarity matrices. A thorough examination of data set similarity matrices can reveal hidden similarities between different data sets. Results show that similarity between different data sets is limited to the type of data and type of analysis.

• Keywords: music, song similarity, recommendation system, text mining, network analysis, vector analysis, matrix similarity, clustering