Automated recommendations can nowadays be found on many online platforms, and such recommendations can create substantial value for consumers and providers. Often, however, not all recommendable items have the same profit margin, and providers might thus be tempted to promote items that maximize their profit. In the short run, consumers might accept non-optimal recommendations, but they may lose their trust in the long run. Ultimately, this leads to the problem of designing balanced recommendation strategies, which consider both consumer and provider value and lead to sustained business success. This work proposes a simulation framework based on Agent-based Modeling designed to help providers explore longitudinal dynamics of different recommendation strategies. In our model, consumer agents receive recommendations from providers, and the perceived quality of the recommendations influences the consumers' trust over time. In addition, we consider network effects where positive and negative experiences are shared with others on social media. Simulations with our framework show that balanced strategies that consider both stakeholders indeed lead to stable consumer trust and sustained profitability. We also find that social media can reinforce phenomena like the loss of trust in the case of negative experiences. To ensure reproducibility and foster future research, we publicly share our flexible simulation framework.
Research on recommender systems algorithms, like other areas of applied machine learning, is largely dominated by efforts to improve the state-of-the-art, typically in terms of accuracy measures. Several recent research works however indicate that the reported improvements over the years sometimes "don't add up", and that methods that were published several years ago often outperform the latest models when evaluated independently. Different factors contribute to this phenomenon, including that some researchers probably often only fine-tune their own models but not the baselines. In this paper, we report the outcomes of an in-depth, systematic, and reproducible comparison of ten collaborative filtering algorithms - covering both traditional and neural models - on several common performance measures on three datasets which are frequently used for evaluation in the recent literature. Our results show that there is no consistent winner across datasets and metrics for the examined top-n recommendation task. Moreover, we find that for none of the accuracy measurements any of the considered neural models led to the best performance. Regarding the performance ranking of algorithms across the measurements, we found that linear models, nearest-neighbor methods, and traditional matrix factorization consistently perform well for the evaluated modest-sized, but commonly-used datasets. Our work shall therefore serve as a guideline for researchers regarding existing baselines to consider in future performance comparisons. Moreover, by providing a set of fine-tuned baseline models for different datasets, we hope that our work helps to establish a common understanding of the state-of-the-art for top-n recommendation tasks.
Recommender systems are software applications that help users find items of interest in situations of information overload in a personalized way, using knowledge about the needs and preferences of individual users. In conversational recommendation approaches, these needs and preferences are acquired by the system in an interactive, multi-turn dialog. A common approach in the literature to drive such dialogs is to incrementally ask users about their preferences regarding desired and undesired item features or regarding individual items. A central research goal in this context is efficiency, evaluated with respect to the number of required interactions until a satisfying item is found. This is usually accomplished by making inferences about the best next question to ask to the user. Today, research on dialog efficiency is almost entirely empirical, aiming to demonstrate, for example, that one strategy for selecting questions is better than another one in a given application. With this work, we complement empirical research with a theoretical, domain-independent model of conversational recommendation. This model, which is designed to cover a range of application scenarios, allows us to investigate the efficiency of conversational approaches in a formal way, in particular with respect to the computational complexity of devising optimal interaction strategies. Through such a theoretical analysis we show that finding an efficient conversational strategy is NP-hard, and in PSPACE in general, but for particular kinds of catalogs the upper bound lowers to POLYLOGSPACE. From a practical point of view, this result implies that catalog characteristics can strongly influence the efficiency of individual conversational strategies and should therefore be considered when designing new strategies. A preliminary empirical analysis on datasets derived from a real-world one aligns with our findings.
Recommender systems are software applications that help users find items of interest in situations of information overload in a personalized way, using knowledge about the needs and preferences of individual users. In conversational recommendation approaches, these needs and preferences are acquired by the system in an interactive, multi-turn dialog. A common approach in the literature to drive such dialogs is to incrementally ask users about their preferences regarding desired and undesired item features or regarding individual items. A central research goal in this context is efficiency, evaluated with respect to the number of required interactions until a satisfying item is found. This is usually accomplished by making inferences about the best next question to ask to the user. Today, research on dialog efficiency is almost entirely empirical, aiming to demonstrate, for example, that one strategy for selecting questions is better than another one in a given application. With this work, we complement empirical research with a theoretical, domain-independent model of conversational recommendation. This model, which is designed to cover a range of application scenarios, allows us to investigate the efficiency of conversational approaches in a formal way, in particular with respect to the computational complexity of devising optimal interaction strategies. Through such a theoretical analysis we show that finding an efficient conversational strategy is NP-hard, and in PSPACE in general, but for particular kinds of catalogs the upper bound lowers to POLYLOGSPACE. From a practical point of view, this result implies that catalog characteristics can strongly influence the efficiency of individual conversational strategies and should therefore be considered when designing new strategies. A preliminary empirical analysis on datasets derived from a real-world one aligns with our findings.
Conversational recommender systems have attracted immense attention recently. The most recent approaches rely on neural models trained on recorded dialogs between humans, implementing an end-to-end learning process. These systems are commonly designed to generate responses given the user's utterances in natural language. One main challenge is that these generated responses both have to be appropriate for the given dialog context and must be grammatically and semantically correct. An alternative to such generation-based approaches is to retrieve responses from pre-recorded dialog data and to adapt them if needed. Such retrieval-based approaches were successfully explored in the context of general conversational systems, but have received limited attention in recent years for CRS. In this work, we re-assess the potential of such approaches and design and evaluate a novel technique for response retrieval and ranking. A user study (N=90) revealed that the responses by our system were on average of higher quality than those of two recent generation-based systems. We furthermore found that the quality ranking of the two generation-based approaches is not aligned with the results from the literature, which points to open methodological questions. Overall, our research underlines that retrieval-based approaches should be considered an alternative or complement to language generation approaches.
Today's research in recommender systems is largely based on experimental designs that are static in a sense that they do not consider potential longitudinal effects of providing recommendations to users. In reality, however, various important and interesting phenomena only emerge or become visible over time, e.g., when a recommender system continuously reinforces the popularity of already successful artists on a music streaming site or when recommendations that aim at profit maximization lead to a loss of consumer trust in the long run. In this paper, we discuss how Agent-Based Modeling and Simulation (ABM) techniques can be used to study such important longitudinal dynamics of recommender systems. To that purpose, we provide an overview of the ABM principles, outline a simulation framework for recommender systems based on the literature, and discuss various practical research questions that can be addressed with such an ABM-based simulation framework.
Recommender systems are designed to help users in situations of information overload. In recent years, we observed increased interest in session-based recommendation scenarios, where the problem is to make item suggestions to users based only on interactions observed in an ongoing session. However, in cases where interactions from previous user sessions are available, the recommendations can be personalized according to the users' long-term preferences, a process called session-aware recommendation. Today, research in this area is scattered and many existing works only compare session-aware with session-based models. This makes it challenging to understand what represents the state-of-the-art. To close this research gap, we benchmarked recent session-aware algorithms against each other and against a number of session-based recommendation algorithms and trivial extensions thereof. Our comparison, to some surprise, revealed that (i) item simple techniques based on nearest neighbors consistently outperform recent neural techniques and that (ii) session-aware models were mostly not better than approaches that do not use long-term preference information. Our work therefore not only points to potential methodological issues where new methods are compared to weak baselines, but also indicates that there remains a huge potential for more sophisticated session-aware recommendation algorithms.
Due to the advances in deep learning, visually-aware recommender systems (RS) have recently attracted increased research interest. Such systems combine collaborative signals with images, usually represented as feature vectors outputted by pre-trained image models. Since item catalogs can be huge, recommendation service providers often rely on images that are supplied by the item providers. In this work, we show that relying on such external sources can make an RS vulnerable to attacks, where the goal of the attacker is to unfairly promote certain pushed items. Specifically, we demonstrate how a new visual attack model can effectively influence the item scores and rankings in a black-box approach, i.e., without knowing the parameters of the model. The main underlying idea is to systematically create small human-imperceptible perturbations of the pushed item image and to devise appropriate gradient approximation methods to incrementally raise the pushed item's score. Experimental evaluations on two datasets show that the novel attack model is effective even when the contribution of the visual features to the overall performance of the recommender system is modest.
In recent years, algorithm research in the area of recommender systems has shifted from matrix factorization techniques and their latent factor models to neural approaches. However, given the proven power of latent factor models, some newer neural approaches incorporate them within more complex network architectures. One specific idea, recently put forward by several researchers, is to consider potential correlations between the latent factors, i.e., embeddings, by applying convolutions over the user-item interaction map. However, contrary to what is claimed in these articles, such interaction maps do not share the properties of images where Convolutional Neural Networks (CNNs) are particularly useful. In this work, we show through analytical considerations and empirical evaluations that the claimed gains reported in the literature cannot be attributed to the ability of CNNs to model embedding correlations, as argued in the original papers. Moreover, additional performance evaluations show that all of the examined recent CNN-based models are outperformed by existing non-neural machine learning techniques or traditional nearest-neighbor approaches. On a more general level, our work points to major methodological issues in recommender systems research.