Toward robust social media sentiment for SMEs: a comparative study of dictionary-based and machine learning approaches with insights for hybrid methodologies

Small and Medium-sized Enterprises (SMEs) increasingly rely on social media to engage customers, promote products, and enhance workplace collaboration. Customer opinions expressed through comments and posts on platforms such as Facebook and Instagram represent valuable insights, yet their informal and context-specific nature—often characterized by slang, misspellings, and bilingual usage—poses challenges for automated sentiment analysis. This study addresses this gap by comparatively evaluating dictionary-based and machine learning approaches to sentiment classification for SMEs' social media content. Data were collected from a diverse set of SMEs across multiple industries, with a substantial volume of customer comments extracted and pre-processed through tokenization, normalization, stop-word removal, and stemming. A customized dictionary was developed to account for local language variations, while Naïve Bayes and Support Vector Machine (SVM) models were employed as supervised classifiers. The findings indicate that dictionary-based methods, while simple and interpretable, struggle with accuracy when processing informal and localized language, whereas machine learning approaches deliver higher overall performance but require extensive preprocessing and tuning. Moreover, the study highlights the potential of hybrid frameworks that combine the interpretability of dictionary-based models with the adaptability of machine learning classifiers. This research contributes both practically and theoretically by (i) demonstrating the limitations of applying generic sentiment analysis tools in localized SME contexts, (ii) proposing a hybrid sentiment analysis framework tailored to SMEs, and (iii) offering empirical evidence to support digital transformation strategies for SMEs in resource-constrained environments. Ultimately, accurate sentiment analysis can enable SMEs to refine business strategies, strengthen customer engagement, and achieve sustainable growth in the digital economy.