Due to privacy and security constraints, directly sharing user data between parties is undesired. Such decentralized data silo issues commonly exist in recommender systems. In general, recommender systems are data-driven. The more data it uses, the better performance it obtains. The data silo issues is a severe limitation of the recommender’s performance. Federated learning is an emerging technology, which bridges the data silos and builds machine learning models without compromising user privacy and data security. We design a recommender system based on federated learning. It is known as the federated recommender system. The system implements plenty of popular algorithms to support various online recommendation services. The algorithm implementation is open-sourced. We also deploy the system on a real-world content recommendation application, achieving significant performance improvement. In this demonstration, we present the architecture of the federated recommender system and give an online demo to show its detailed working procedures and results in content recommendations.
Realistic recommender systems are often required to adapt to ever-changing data and tasks or to explore different models systematically. To address the need, we present AutoRec 1 2, an open-source automated machine learning (AutoML) platform extended from the TensorFlow [3] ecosystem and, to our knowledge, the first framework to leverage AutoML for model search and hyperparameter tuning in deep recommendation models. AutoRec also supports a highly flexible pipeline that accommodates both sparse and dense inputs, rating prediction and click-through rate (CTR) prediction tasks, and an array of recommendation models. Lastly, AutoRec provides a simple, user-friendly API. Experiments conducted on the benchmark datasets reveal AutoRec is reliable and can identify models which resemble the best model without prior knowledge.
We introduce Auto-Surprise1, an automated recommender system library. Auto-Surprise is an extension of the Surprise recommender system library and eases the algorithm selection and configuration process. Compared to an out-of-the-box Surprise library, without hyper parameter optimization, AutoSurprise performs better, when evaluated with MovieLens, Book Crossing and Jester datasets. It may also result in the selection of an algorithm with significantly lower runtime. Compared to Surprise’s grid search, Auto-Surprise performs equally well or slightly better in terms of RMSE, and is notably faster in finding the optimum hyperparameters.
The field of recommender systems has rapidly evolved over the last few years, with significant advances made due to the in-flux of deep learning techniques. However, as a result of this rapid progress, escalating barriers-to-entry for new researchers is emerging. In particular, state-of-the-art approaches have fragmented into a large number of code-bases, often requiring different input formats, pre-processing stages and evaluating with different metric packages. Hence, it is time-consuming for new researchers to reach the point of having both an effective baseline set and a sound comparative environment. As a step towards elevating this problem, we have developed BETA-Rec, an open source project for Building, Evaluating and Tuning Automated Recommender Systems. BETA-Rec aims to provide a practical data toolkit for building end-to-end recommendation systems in a standardized way. It provides means for dataset preparation and splitting using common strategies, a generalized model engine for implementing recommender models using Pytorch with 9 models available out-of-the-box, as well as a unified training, validation, tuning and testing pipeline. Furthermore, BETA-Rec is designed to be both modular and extensible, enabling new models to be quickly added to the framework. It is deployable in a wide range of environments via pre-built docker containers and supports distributed parameter tuning using Ray. In this demo, we will illustrate the deployment and use of BETA-Rec for researchers and practitioners on a number of standard recommendation datasets. The source code of the project is available at github: https://github.com/beta-team/beta-recsys.
We develop RecSim NG, a probabilistic platform that supports natural, concise specification and learning of models for multi-agent recommender systems simulation. RecSim NG is a scalable, modular, differentiable simulator implemented in Edward2 and TensorFlow.
Comparative experimentation is important for studying reproducibility in recommender systems. This is particularly true in areas without well-established methodologies, such as fairness-aware recommendation. In this paper, we describe fairness-aware enhancements to our recommender systems experimentation tool librec-auto. These enhancements include metrics for various classes of fairness definitions, extension of the experimental model to support result re-ranking and a library of associated re-ranking algorithms, and additional support for experiment automation and reporting. The associated demo will help attendees move quickly to configuring and running their own experiments with librec-auto.
We present PicTouRe – a picture-based tourism recommender. PicTouRe aims to mitigate people’s difficulties in explicitly expressing their touristic preferences, which is even more challenging in the initial phase of travel decision making. Addressing this issue, with PicTouRe we follow the idiom “a picture is worth a thousand words” and use pictures as a tool to implicitly elicit peoples’ touristic preferences. We describe the core concept of PicTouRe – the Generic Profiler, which in essence determines an explainable vector representation, i.e., touristic profile, given any picture collection as input. We showcase a user’s journey through PicTouRe and describe the steps behind. Finally, we present results of a first user study supporting our approach. PicTouRe is available under https://pictoprof.ec.tuwien.ac.at and a demo video under https://youtu.be/xZnXLPcenEs.
“We introduce Recommender-Systems.com (RS_c) as a central platform for the recommender-systems community. RS_c provides regular news on important events in the community as well as curated lists of recommender-system resources including datasets, algorithms, jobs, software, and learning materials. Based on a survey with 28 participants – mostly authors at the RecSys 2019 conference – 91% agree that RS_c could be a major contribution to the community. Participants consider it currently particularly difficult to find best practice guidelines (45%); researchers, freelancers and employers (45%); and curated lists of state-of-the-art algorithms, software, and datasets (36%). Notably, only 19% consider it (very) easy to find material relating to diversity, equality and anti-discrimination.”
Hybrid recommender systems achieve state-of-the-art performance by integrating several different information sources along with multiple recommendation approaches. Probabilistic Soft Logic (PSL) has been shown to be an accessible and effective means of creating extensible hybrid recommenders [11]. PSL allows users to easily create intuitive models that incorporate background information and capture complex interactions. However these complex interactions can sometimes make PSL models difficult to inspect, debug, and understand. In this paper, we present a generic visual model inspector for PSL, and show how our inspector can be used on a hybrid recommender system to: debug errors in the model, analyze the performance of individual components of the model, and explain recommendations made by the model.
Recommender systems are key tools to push the consumption of items. Imbalances in the data distribution might affect the exposure given to providers, thus affecting their business and experience in the platform. In my work, I study the impact of data imbalances for the stakeholders, according to how recommendations are generated.
In my Ph.D. work, my objective is to improve the state of the art in Conversational Recommender Systems, by proposing a model that closely follows the process that people enact when searching for products and services. Rich user profiles are elicited using natural language dialogue. Item descriptions will be extracted from a combination of structured and unstructured data such as user reviews. Natural language explanations will ensure that users can quickly understand the reasoning behind the recommendations. Interactive explanation will then allow them to further compare several alternatives. This extended abstract presents the motivations of my work, it details the research plan, and the research questions. Finally, it shows some preliminary results, and outlines the next steps for my Ph.D. program.
Broadly, the goal of my research is to develop modeling techniques for recommender systems data in the streaming context. Streaming models in recommender systems have received attention in demanding applications such as social media streams and news delivery as model requirements are complex. To date, my work in recommender systems has focused on collaborative filtering with work in exploiting structure in similarity evaluation, modeling event time of user activity, and confidence measures for individual predictions. Exploiting underlying structure in similarity evaluation is shown to positively impact prediction accuracy. Modeling event times of user activity presents many applications in soliciting user involvement in ecommerce settings. Early work is in progress in developing confidence measures for individual rating predictions. While these efforts have not been focused on the streaming environment to date, they do form a basis for future work in which computational demands are higher. Extensions to the streaming domain are discussed.
Throughout the years, numerous recommendation algorithms have been developed to address the information filtering problem by leveraging users’ tastes through implicit or explicit feedback. In this paper, we present the work undertaken as part of a PhD thesis focused on exploring new evaluation dimensions centred around the efficiency-effectiveness trade-offs present in state-ofthe-art recommendation systems. Firstly, we highlight the lack of efficiency-oriented studies and we formulate the research problem. Then, we propose a mapping of the design space and a classification of the recommendation algorithms/models with respect to salient attributes and characteristics. At the same time, we explain why and how assessing the recommendations on an accuracy versus training cost curve would advance the current knowledge in the area of evaluation, as well as open new research avenues for exploring parameter configurations within well-known algorithms. Finally, we make the case for a comprehensive methodology that incorporates predictive efficiency-effectiveness models, which illustrate the performance and behaviour of the recommendation systems under different recommendation tasks, while satisfying user-defined quality of service constraints and goals.
“Tasks as physical training planning, computer hardware configuration or fully dietary advice, are problems that exhibit multiple choices, composed in turn of simpler items (specific exercises, components or recipes). An ideal recommender system would not only recommend simple items based on the user’s tastes, but would offer a set of items that suit the user’s needs and preferences so that they form a meaningful structure that can evolve in time. Taking this idea as our main cornerstone, this Ph.D. face two objectives: being able to generate an item with a complex structure from simpler items, and integrating the user’s limitations and preferences to develop an adaptive recommender system. This work proposes the incorporation of item generation systems as a preprocessing stage in the recommendation process through evolutionary algorithms. We found that genetic algorithms are still an interesting and powerful computational tool that have not been fully developed for this task. We a theoretical model that can deal with problems from different areas, incorporating the ability to recommend complex objects with flexible restrictions and specific structures that depend on the user an its evolution. Finally this approach will be tested in different scenarios, as Stance4Health European project, where a personalized nutrition service based on user microbiota is being developed based on this concept.”
A common and recently widely accepted problem in the field of machine learning is the black box nature of many algorithms. In practice, many machine learning algorithms can only be viewed and evaluated in terms of their inputs and outputs, without taking their internal workings into account. Perhaps the most notorious examples in this context are artificial neural networks and deep learning techniques, but they are certainly not the only techniques that suffer from this problem. Matrix factorisation models for recommendation systems, for example, suffer from the same lack of interpretability. Our research focuses on applying and adapting pattern mining techniques to gain meaningful insights in recommendation algorithms by analysing them in terms of both their input and output, also allowing us to compare different algorithms and discover the hidden biases that lead to those differences.
Traveling for leisure has become an important part of our society. It has proven time and again its benefits for wellbeing and personal growth. There are many types of tourism and one of them is Accessible Tourism (AT), an ongoing endeavor to ensure that everyone, regardless of condition, has the right to benefit from tourism experiences. Recommender systems (RSs) represent a mature technique for generating clear and personalized suggestions. While being widely researched and used by the tourism academic community and the tourism industry in general, Recommender Systems (RSs) can still do much more for Accessible Tourism (AT). This thesis aims to build a recommender system dedicated to recommending accessible tourism destinations and easy the process of e2e trip planning for people with disabilities. With a modular design, use of ontologies, machine learning techniques and a “start small, define expansion, expand” approach, this recommender system, once built, aims to be validated by real users
Traditionally, recommender systems are built on the users’ item consumption history. Many times, the users also make items reviews, giving us additional information in the form of text and images that are generally, not fully exploited. In this research we propose different approaches in the context of restaurants recommendation, where we take advantage of this information, by extracting a semantic meaning, in order to improve the traditional RS in terms of personalization and explainability
