Proceedings of the International scientific and practical conference ―Oxford International Science Forum‖ (April 17-19, 2026) / Publisher website: www.naukainfo.com. - Oxford, United Kingdom, 2026. - 367 p.

104 accessibility, but also data semantics, campaign logic, and the meaning of outcome variables [7-11]. Challenges of CD collection . One of the most serious problems in crowdfunding research is the incomplete access to relevant data. Even when campaign pages are public, detailed user interaction histories, complete support logs, or information about specific backers are often not available to researchers [7-8, 10]. A second problem is the heterogeneity of data structures and values across platforms. Similar fields may have different interpretations depending on whether the platform is reward-based, donation-based, or share-based. As a result, cross-platform datasets require additional special processing or semantic harmonization before they can be meaningfully analyzed [2, 9-11]. The third problem is multimodality. Crowdfunding datasets often combine numeric metrics, text, timestamps, categories, and interaction traces. Integrating them into a single analytical representation requires extensive preprocessing, including cleaning, normalization, temporal aggregation, feature engineering, and natural language processing (NLP) [4-6]. A fourth problem is bias and incompleteness. Available data may overrepresent prominent or successful campaigns, while underrepresenting unsuccessful, remote, obscure, or historically controversial campaigns. This creates survivorship bias and affects the validity of analytical conclusions [1-3]. The fifth issue is related to ethical and legal constraints. Data collection through parsing or indirect collection must take into account platform terms, privacy considerations, and acceptable uses of user-generated content such as comments, follower tracks, and attributes associated with identification [7-10]. Thus, CD collection should be considered as a methodological stage of the research itself, rather than as a purely technical preliminary step [2], [11-13]. Methods and tools for CD mining . Different types of data analytics require different classes of analytical methods and tools. For identifying success factors, suitable methods include descriptive statistics, correlation analysis, logistic regression, decision trees, random forests, and gradient