Consumer profiling has become one of the most common practices on online trading platforms. Many platforms strive to obtain and implement technological innovations that allow them to understand and identify consumers’ needs, and, thereafter, monetize this capability by charging sellers to present and/ or sell their products or services based on consumers’ interests. However, an interesting and relevant question arises in this context: Does the platform have an incentive to profile its buyers as accurately as possible? This paper develops and analyzes a parsimonious game-theoretic model to answer this research question. We find that, surprisingly, platforms that charge sellers for discoveries have a perverse incentive to deviate from accurate buyer profiling. However, such a perverse incentive does not exist for platforms that charge sellers for transactions. As a result, with such a perverse incentive, social welfare under discovery-based pricing is lower than that under transaction-based pricing.

Identifying Perverse Incentives in Buyer Profiling on Online Trading Platforms

In present scenario, to improve the consumers buying/purchasing experience, use of technological innovations like Recommender system is very prominent. This small yet powerful utility analyzes the buying pattern of a consumer and suggests which items to buy or use. Recommender systems can be applied to varied fields of consumer’s interest like online shopping, ticket booking and other online contents. Most of the e-commerce sites are attracting huge number of potential customers by providing useful suggestions regarding buying a product or service. This technique is mainly based on the use of Machine Learning, which enables the system to make decisions efficiently. Earlier, these recommendations were mainly relying on filtering process such as collaborative filtering, content based, knowledge based, demographic and hybrid filtering. These filtering techniques, which constitute the recommender system, are discussed in detail in this study. The survey is conducted to describe the various means and methods of recommendation to consumer in real time. This survey presents a comparative study among different types of recommender system based on various parameters and filtering schemes. Moreover, it shows a significant improvement in recommender system by using machine learning based approaches.

R. Rajkumar

Papers

A Comprehensive Survey on Web Recommendations Systems with Special Focus on Filtering Techniques and Usage of Machine Learning