Automated Workflow Recommendations in Orange by Learning from a Collection of Workflow Screenshots

Eltsova, Maria (author), Zupan, Blaž (mentor)

This thesis presents a recommender system that suggests relevant widgets for Orange Data Mining workflows by learning from a curated collection of workflow screenshots. The dataset consists of 819 screenshots sourced from 25 chapters of Data Mining tutorials, created by various authors over time. We detect widgets in these screenshots using a two-stage computer vision pipeline: the Circle Hough Transform locates circular widget regions, and the Scale Invariant Feature Transform (SIFT) matches these regions against the Orange Widget Catalogue. The detection results — counts of each widget present in each screenshot — form a widget–screenshot matrix, which serves as the input for collaborative filtering. We train a Biased Regularized Incremental Simultaneous Matrix Factorization (BRISMF) model to predict the most suitable next widget for a partially built workflow. In evaluation, the BRISMF-based recommender achieved an average reconstructed position of 5.04, outperforming a frequency-based baseline (5.76) by 12.5%. These findings suggest the potential for automated recommendations in workflow design within visual programming environments.

• Language: English
• Keywords: Orange Data Mining, Houghova transformacija, transformacijaznačilk z invarianco glede na merilo, pristranska regularizirana inkrementalnasimultana matriˇcna faktorizacija
• Work type: Bachelor thesis/paper
• Typology: 2.11 - Undergraduate Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2025
• PID: 20.500.12556/RUL-175501
• COBISS.SI-ID: 257364483
• Publication date in RUL: 29.10.2025

Secondary language

• Language: Slovenian
• Title: Priporočila delotokov za Orange z učenjem iz zbirke posnetkov zaslonov delotokov

Abstract

V diplomskem delu je predstavljen priporočilni sistem za predlaganje ustreznih komponent v delotokih okolja Orange DataMining na podlagi analize obstoječih vzorcev. Sistem preverja možnost samodejnega prepoznavanja logiˇcnih naslednjih korakov v še nedokončanih delotokih. Vhodni podatki vsebujejo 819 posnetkov zaslona iz 25 poglavij teorije podatkovnega rudarjenja. Komponente smo zaznavali s Houghovo transformacijo za kroge in transformacijo značilk z invarianco glede na merilo (metoda SIFT), pri čemer so rezultati zaznavanja oblikovali matriko komponenta–posnetek za kolaborativno filtriranje. Priporočilni model temelji na pristranski regularizirani inkrementalni simultani matrični faktorizaciji (BRISMF). Vrednotenje je pokazalo 12,5% boljšo uspešnost v primerjavi z osnovnim modelom, ki temelji na pogostosti, kar potrjuje potencial avtomatiziranih priporočil pri načrtovanju delotokov v okoljih za vizualno programiranje.

• Keywords: Orange Data Mining, Hough Transform, Scale Invariant FeatureTransform, Biased Regularized Incremental Simultaneous Matrix Factorization