graduate student from 01.01.2025 until now
Elektrostal', Moscow, Russian Federation
VAK Russia 5.2.3
VAK Russia 5.2.4
VAK Russia 5.2.5
VAK Russia 5.2.6
VAK Russia 5.2.7
UDC 330.42
Consumer preferences in digital product markets change faster than traditional demand monitoring tools can capture. Digital platforms accumulate thousands of user reviews every day, which may contain early signs of such changes. However, most studies analyze reviews only in terms of sentiment and volume, while overlooking the topic structure of user content and its dynamics over time. This paper proposes a methodology for identifying early signals of demand changes through the analysis of topic trends in user-generated content. Using a corpus of Steam user reviews, an LDA topic model was constructed, the dynamics of topic shares across time windows were tracked, and the leading nature of topic-based indicators was tested against demand proxies. Granger causality tests and cross-correlation analysis were used to assess the statistical relationship. Incorporating topic-based indicators into an ARIMAX model did not improve forecasting accuracy compared with the baseline ARIMA model, which indicates the need for further development of the forecasting toolkit. The results show that changes in the topical structure of reviews can serve as leading indicators of demand and can be used to build early warning systems for business.
user-generated content, topic modeling, time series, demand proxies, cross-correlation, forecasting
1. Il'chenko, P. V. Vliyanie ugc-kontenta na povedenie potrebiteley i prinyatie resheniy o pokupke // Ekonomika i biznes: teoriya i praktika. 2024. №4-2 (110). URL: https://cyberleninka.ru/article/n/vliyanie-ugc-kontenta-na-povedenie-potrebiteley-i-prinyatie-resheniy-o-pokupke.
2. Zaharova, A. A. Ocenka tonal'nosti kommentariev pol'zovateley metodami tematicheskogo modelirovaniya / A. A. Zaharova // Sbornik trudov IX Kongressa molodyh uchenyh, Sankt-Peterburg, 15–18 aprelya 2020 goda. Tom 1. – Universitet ITMO, Sankt-Peterburg: federal'noe gosudarstvennoe avtonomnoe obrazovatel'noe uchrezhdenie vysshego obrazovaniya "Nacional'nyy issledovatel'skiy universitet ITMO", 2021. – S. 207-211. – EDN UDUBLN.
3. Semenova, E. (2025) «Funkcional'nye osobennosti UGC na primere setevyh media Tambovskoy oblasti», Dinamika mediasistem, 5(2), ss. 210–218. doi:https://doi.org/10.47475/2949-3390-2025-5-2-210-218.
4. Nestrukturirovannye dannye: primery, instrumenty, metodiki i rekomendacii [Elektronnyy resurs] // Habr. URL: https://habr.com/ru/articles/756454.
5. Voytik, U. V., Sidorova, N. I. Pol'zovatel'skiy kontent (UGC) v marketinge Biznes-pul's : II Mezhdunar. nauch.-prakt. stud. konf., Minsk, 10 noyabrya 2023 g. : sb. materialov / redkol.: V. V. Mancurova [i dr.]. – Minsk : Institut biznesa BGU, 2024. – S. 221-224.
6. Steam Reviews Dataset [Elektronnyy resurs] // Kaggle. URL: https://www.kaggle.com/datasets/forgemaster/steam-reviews-dataset/code.
