Influencer Recommendation from Campaign Briefings: A Briefing-Centric Approach Using Sparse Representations and Density-Based Clustering

• Luiz Carlos Grandisoli UPM
• Gustavo Scalabrini Sampaio UPM
• Maria Amelia Eliseo UPM

DOI: https://doi.org/10.5753/eniac.2026.29713

Resumo

This work evaluates a cold-start influencer recommendation pipeline using 110 synthetic campaign briefings and 1,100 synthetic influencer biographies in Brazilian Portuguese. The method compares TF-IDF and multilingual MiniLM embeddings under direct ranking and clustering-based retrieval. Results show that TF-IDF combined with DBSCAN achieves the best performance, reaching P@5 = 0.844 and NDCG@5 = 0.854 overall, and P@5 = 0.928 in the in-scope evaluation. The findings indicate that briefing-centric recommendation is feasible in cold-start settings, and that within this synthetic benchmark, where niche labels are defined by characteristic vocabulary, sparse lexical representations outperform dense embeddings for niche-level discrimination; the advantage persists under nested cross-validation (p < 0.001) but is specific to lexically marked niches rather than general.

Palavras-chave: Recommendation systems, natural language processing, influencer-based recommendation, text clustering, semantic similarity, content-based recommendation

Referências

Sweet, T., Rothwell, A. and Luo, X. (2019). Machine learning techniques for brand-influencer matchmaking on the Instagram social network. arXiv preprint arXiv:1906.05026.

Jha, A. K. and Ray, S. (2023). Influencer recommendation system: choosing the right influencer using a network analysis approach. Marketing Intelligence & Planning, 41(3):369–384.

Gan, T., Wang, S., Liu, M., Song, X., Yao, Y. and Nie, L. (2019). Seeking micro-influencers for brand promotion. In Proceedings of the 27th ACM International Conference on Multimedia, pages 1595–1603. ACM.

Wang, S., Gan, T., Liu, Y., Wu, J., Cheng, Y. and Nie, L. (2023). Micro-influencer recommendation by multi-perspective account representation learning. IEEE Transactions on Multimedia, 25:7463–7475.

Zhang, S., Yao, L., Sun, A. and Tay, Y. (2019). Deep learning based recommender system: a survey and new perspectives. ACM Computing Surveys, 52(1):1–38.

Stefanovic, P., Kurasova, O. and Strimaitis, R. (2019). The N-Grams based text similarity detection approach using self-organizing maps and similarity measures. Applied Sciences, 9(9):1870.

Cretulescu, R., Morariu, D. and Vintan, L. (2011). Clustering text documents: an overview. Acta Universitatis Cibiniensis, Technical Series.

Callaghan, S., Doner, H., Medalsy, J., Pione, A. and Teichner, W. (2024). The trends defining the global wellness market in 2024. McKinsey & Company.

Global Wellness Institute (2024). Global wellness economy monitor 2024. Global Wellness Institute.

Fuchs, M. and Hopken, W. (2022). Clustering: hierarchical, k-Means, DBSCAN. In Egger, R., editor, Applied Data Science in Tourism, pages 129–130. Springer.

Manning, C. D., Raghavan, P. and Schutze, H. (2008). Introduction to Information Retrieval. Cambridge University Press.

Aggarwal, C. C. (2016). Recommender Systems: The Textbook. Springer.

Reimers, N. and Gurevych, I. (2019). Sentence-BERT: sentence embeddings using Siamese BERT-networks. In Proceedings of EMNLP-IJCNLP, pages 3982–3992. Association for Computational Linguistics.

Thakur, N., Reimers, N., Rücklé, A., Srivastava, A. and Gurevych, I. (2021). BEIR: a heterogeneous benchmark for zero-shot evaluation of information retrieval models. In Proceedings of NeurIPS Datasets and Benchmarks.

Grandisoli, L. C., Scalabrini Sampaio, G. and Eliseo, M. A. (2026a). A synthetic Brazilian Portuguese benchmark for briefing-centric influencer recommendation [Data set]. Zenodo. DOI: 10.5281/zenodo.22047229.

Grandisoli, L. C., Scalabrini Sampaio, G. and Eliseo, M. A. (2026b). Reproduction code for influencer recommendation from campaign briefings [Software]. Zenodo. DOI: 10.5281/zenodo.22048849.