Identifying influencer market manipulations and recommending engaging accounts

Official URL: https://risc01.sabanciuniv.edu/record=b3147941_(Table of contents)

Today, as consumers are channeled to social media platforms, the demands of companies and brands to advertise and promote on social media platforms are constantly increasing. Companies and brands that have been searching for different advertising and promotion approaches use influencers. They make an agreement with the influencer on a fee per story or per post to promote their products. The increasing number of social media users has led to the growth of the race within the influencer market. Some influencers have begun to use various methods that boost their engagement metrics artificially. Purchasing bot followers or automated engagement are examples of such manipulative efforts. As a result of this, although the engagement numbers of influencer seem high, they have blurred the organic engagement rate and misled the companies that hire influencers. In this thesis, we present a new metric, the CRE (capture-recapture engagement) score, to the literature that can detect organic interactions more accurately than existing interaction metrics used in influencer marketing agencies. As a result of the evaluations made, it has been observed that the metric we presented offers better performance than the metrics used in the literature. In addition to this, we introduce an influencer recommendation system built by using the CRE score. The proposed system can identify influencers that have higher engagements while preserving the similarity of the profile content with the target user. This approach provides opportunities to select highly engaging but less popular influencers.