Hyppää sisältöön
    • Suomeksi
    • På svenska
    • In English
  • Suomi
  • Svenska
  • English
  • Kirjaudu
Hakuohjeet
JavaScript is disabled for your browser. Some features of this site may not work without it.
Näytä viite 
  •   Ammattikorkeakoulut
  • Hämeen ammattikorkeakoulu
  • Opinnäytetyöt
  • Näytä viite
  •   Ammattikorkeakoulut
  • Hämeen ammattikorkeakoulu
  • Opinnäytetyöt
  • Näytä viite

Automated Tone Analysis of Websites for Kalibro.io

Rankiri Pathirage, Anuradha Priyankara (2025)

 
Avaa tiedosto
Rankiri_Pathirage_Anuradha_Priyankara.pdf (1.665Mt)
Lataukset: 


Rankiri Pathirage, Anuradha Priyankara
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Näytä kaikki kuvailutiedot
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025090124320
Tiivistelmä
The purpose of this thesis was to develop an AI-powered system for automated tone analysis of website content for Kalibro.io, a digital optimization platform by Calevala Interactive Ltd. While Kalibro.io currently evaluates technical aspects like mobile friendliness, it does not yet analyse complex content tones. This process is not handled manually either. To address this gap, the research aimed to support the development of a new AI tool for Kalibro.io by extracting and segmenting text from website screenshots into meaningful chunks and classifying each segment into one of ten tone categories (Positive, Negative, Neutral, Formal, Informal, Confident, Friendly, Aggressive, Urgent, Informative). Key research questions included that How accurately can vision-language models extract text from website screenshots? And can fine-tuned NLP models reliably classify tones across diverse content types? The thesis adopted a practical, iterative development approach. First, text extraction was implemented using PaliGemma 2, a vision-language model, followed by segmentation via Open AI's GPT 4.1 to isolate logical units (e.g., headlines, paragraphs). For tone classification, RoBERTa was fine-tuned on a custom dataset labelled with the ten target tones. The system was tested on real-world websites, with performance evaluated using metrics like accuracy and F1-score, supplemented by qualitative case.
This project introduces an innovative AI-powered tone analysis capability to Kalibro.io's platform, expanding its competitive advantage in digital optimization. By automatically evaluating content tone across ten distinct categories, the solution adds a new dimension to website assessment that complements existing technical metrics. The implementation enables Kalibro.io to offer clients unique insights into brand voice alignment and audience engagement - previously unavailable through conventional analysis tools. This enhancement positions the platform to deliver more comprehensive digital strategy services while maintaining its scalable, automated approach to website evaluation.
Kokoelmat
  • Opinnäytetyöt
Ammattikorkeakoulujen opinnäytetyöt ja julkaisut
Yhteydenotto | Tietoa käyttöoikeuksista | Tietosuojailmoitus | Saavutettavuusseloste
 

Selaa kokoelmaa

NimekkeetTekijätJulkaisuajatKoulutusalatAsiasanatUusimmatKokoelmat

Henkilökunnalle

Ammattikorkeakoulujen opinnäytetyöt ja julkaisut
Yhteydenotto | Tietoa käyttöoikeuksista | Tietosuojailmoitus | Saavutettavuusseloste