Semantička segmentacija teksta računa

undergraduate thesis

• Title (english): Semantic Segmentaion of Invoice Text
• Author: Dominik Čubelić
• Mentor: Jan Šnajder (mentor)
• Committee member: Jan Šnajder (predsjednik povjerenstva)
• Committee member: Martin Tutek (član povjerenstva)
• Committee member: Zoran Medić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu su se promatrali modeli semantičke segmentacije teksta računa. Riječi očitane s računa se klasificiraju po tri semantičke značajke: proizvod, cijena i ostalo. Razvijena su tri modela: HMM, CRF i LSTM. Najbolje rezultate ostvario je CRF s F1 rezultatom preko 90\% što je uvelike nadmašilo referentni model temeljen na heuristici.

Abstract (english)

In this thesis, semantic segmentation of invoice text models was observed. Words extracted from invoices are classified on three semantic labels: product, price and other. Three models were developed: HMM, CRF and LSTM. Best results were achieved by CRF with F1 score over 90\% which greatly surpassed the referent model based on heuristics.

• Keywords: strojno učenje; obrada prirodnog jezika; HMM; CRF; LSTM; označavanje sljedova; računi
• Keywords (english): machine learning; natural language processing; HMM; CRF; LSTM; sequence labelling; invoices
• Language: croatian
• URN:NBN: urn:nbn:hr:168:787814
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 17:23:22