Connecting Social Media to E Commerce for Cold Start Product Recommendation using Microblogging Information
Mekala Vamshi Krishna
Information Technology, Vardhaman College
of Engineering, Hyderabad, India
vamshikrishnaa121@gmail.com
P Abhilash
Information Technology, Vardhaman College of
Engineering, Hyderabad, India
pulledulaabhilash24@gmail.com
Kola Shashivardhan
Information Technology, Vardhaman College
of Engineering, Hyderabad, India
kolashashivardhan40@gmail.co m
Gangoni Shreshta, Mannava Yesubabu
Information Technology,
Vardhaman College of Engineering, Hyderabad, India
gangonishreshta324@gmail.com
mannavababu@gmail.com
Abstract— The rapid expansion of social media has reshaped the way consumers discover and evaluate products, creating a continuous stream of opinions, preferences, and behavioural cues. However, most e-commerce platforms still struggle to recommend newly launched items because such products lack historical interactions, ratings, and user feedback. This challenge, known as the cold start problem, limits the accuracy and relevance of traditional recommendation models. The present study explores an integrated approach that links microblogging platforms with e-commerce environments to generate effective recommendations for new products. The work examines how short posts, hashtags, user sentiments, and emerging discussion trends can be transformed into meaningful indicators of user interest. A tailored pipeline is developed to collect microblogging content, filter domain-specific expressions, extract feature patterns, and map them to product attributes in an e-commerce catalogue. The proposed method aims to bridge early-stage information gaps by identifying consumer intent signals before a product accumulates transactional data. Experimental evaluation demonstrates that incorporating microblogging information reduces sparsity, enhances the visibility of new items, and improves the overall recommendation relevance compared to baseline methods relying solely on e-commerce behaviour. The findings confirm that social micro-interactions can serve as reliable external knowledge for alleviating cold start limitations. This study contributes an adaptable framework that enables businesses to respond more quickly to emerging product interests and to offer personalised suggestions from the moment a product enters the market.
Keywords— Microblogging Analysis; Cold-Start Recommendation; Social Media Integration; E-commerce Personalisation; Sentiment and Topic Modelling; Product Visibility Enhancement.
