hh.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
A hybrid dependency-based approach for Urdu sentiment analysis
Capital University of Science and Technology, Islamabad, Pakistan.
Halmstad University, School of Information Technology. Royal Institute of Technology, Stockholm, Sweden.ORCID iD: 0000-0002-8933-7894
King Saud University, Riyadh, Saudi Arabia.ORCID iD: 0000-0001-7191-2099
Saudi Electronic University, Riyadh, Saudi Arabia.
Show others and affiliations
2023 (English)In: Scientific Reports, E-ISSN 2045-2322, Vol. 13, article id 22075Article in journal (Refereed) Published
Abstract [en]

In the digital age, social media has emerged as a significant platform, generating a vast amount of raw data daily. This data reflects the opinions of individuals from diverse backgrounds, races, cultures, and age groups, spanning a wide range of topics. Businesses can leverage this data to extract valuable insights, improve their services, and effectively reach a broader audience based on users’ expressed opinions on social media platforms. To harness the potential of this extensive and unstructured data, a deep understanding of Natural Language Processing (NLP) is crucial. Existing approaches for sentiment analysis (SA) often rely on word co-occurrence frequencies, which prove inefficient in practical scenarios. Identifying this research gap, this paper presents a framework for concept-level sentiment analysis, aiming to enhance the accuracy of sentiment analysis (SA). A comprehensive Urdu language dataset was constructed by collecting data from YouTube, consisting of various talks and reviews on topics such as movies, politics, and commercial products. The dataset was further enriched by incorporating language rules and Deep Neural Networks (DNN) to optimize polarity detection. For sentiment analysis, the proposed framework employs predefined rules to trigger sentiment flow from words to concepts, leveraging the dependency relations among different words in a sentence based on Urdu language grammatical rules. In cases where predefined patterns are not triggered, the framework seamlessly switches to its sub-symbolic counterpart, passing the data to the DNN for sentence classification. Experimental results demonstrate that the proposed framework surpasses state-of-the-art approaches, including LSTM, CNN, SVM, LR, and MLP, achieving an improvement of 6–7% on Urdu dataset. In conclusion, this research paper introduces a novel framework for concept-level sentiment analysis of Urdu language data sourced from social media platforms. By combining language rules and DNN, the proposed framework demonstrates superior performance compared to existing methodologies, showcasing its effectiveness in accurately analyzing sentiment in Urdu text data. © 2023, The Author(s).

Place, publisher, year, edition, pages
London: Nature Publishing Group, 2023. Vol. 13, article id 22075
National Category
Natural Language Processing
Identifiers
URN: urn:nbn:se:hh:diva-52319DOI: 10.1038/s41598-023-48817-8PubMedID: 38086933Scopus ID: 2-s2.0-85179644602OAI: oai:DiVA.org:hh-52319DiVA, id: diva2:1822494
Funder
KTH Royal Institute of Technology
Note

Funding: Open access funding provided by Royal Institute of Technology

Available from: 2023-12-22 Created: 2023-12-22 Last updated: 2025-02-07Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textPubMedScopus

Authority records

Kanwal, Summrina

Search in DiVA

By author/editor
Kanwal, SummrinaAllheeib, Nasser I.Gogate, MandarKhashan, Osama A.
By organisation
School of Information Technology
In the same journal
Scientific Reports
Natural Language Processing

Search outside of DiVA

GoogleGoogle Scholar

doi
pubmed
urn-nbn

Altmetric score

doi
pubmed
urn-nbn
Total: 14 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf