There has been great interest in fairness in machine learning, especially in relation to classification problems. In ranking-related problems, such as in online advertising, recommender systems, and HR automation, much work on fairness remains to be done. Two complications arise: first, the protected attribute may not be available in many applications. Second, there are multiple measures of fairness of rankings, and optimization-based methods utilizing a single measure of fairness of rankings may produce rankings that are unfair with respect to other measures. In this work, we propose a randomized method for post-processing rankings, which do not require the availability of the protected attribute. In an extensive numerical study, we show the robustness of our methods with respect to P-Fairness and effectiveness with respect to Normalized Discounted Cumulative Gain (NDCG) from the baseline ranking, improving on previously proposed methods.

Collaborative filtering (CF) has exhibited prominent results for recommender systems and been broadly utilized for real-world applications. A branch of research enhances CF methods by message passing used in graph neural networks, due to its strong capabilities of extracting knowledge from graph-structured data, like user-item bipartite graphs that naturally exist in CF. They assume that message passing helps CF methods in a manner akin to its benefits for graph-based learning tasks in general. However, even though message passing empirically improves CF, whether or not this assumption is correct still needs verification. To address this gap, we formally investigate why message passing helps CF from multiple perspectives and show that many assumptions made by previous works are not entirely accurate. With our curated ablation studies and theoretical analyses, we discover that (1) message passing improves the CF performance primarily by additional representations passed from neighbors during the forward pass instead of additional gradient updates to neighbor representations during the model back-propagation and (ii) message passing usually helps low-degree nodes more than high-degree nodes. Utilizing these novel findings, we present Test-time Aggregation for CF, namely TAG-CF, a test-time augmentation framework that only conducts message passing once at inference time. The key novelty of TAG-CF is that it effectively utilizes graph knowledge while circumventing most of notorious computational overheads of message passing. Besides, TAG-CF is extremely versatile can be used as a plug-and-play module to enhance representations trained by different CF supervision signals. Evaluated on six datasets, TAG-CF consistently improves the recommendation performance of CF methods without graph by up to 39.2% on cold users and 31.7% on all users, with little to no extra computational overheads.

In complex financial markets, recommender systems can play a crucial role in empowering individuals to make informed decisions. Existing studies predominantly focus on price prediction, but even the most sophisticated models cannot accurately predict stock prices. Also, many studies show that most individual investors do not follow established investment theories because they have their own preferences. Hence, the tricky point in stock recommendation is that recommendations should give good investment performance but also should not ignore individual preferences. To develop effective stock recommender systems, it is essential to consider three key aspects: 1) individual preferences, 2) portfolio diversification, and 3) temporal aspect of both stock features and individual preferences. In response, we develop the portfolio temporal graph network recommender PfoTGNRec, which can handle time-varying collaborative signals and incorporates diversification-enhancing contrastive learning. As a result, our model demonstrated superior performance compared to various baselines, including cutting-edge dynamic embedding models and existing stock recommendation models, in a sense that our model exhibited good investment performance while maintaining competitive in capturing individual preferences. The source code and data are available at https://anonymous.4open.science/r/IJCAI2024-12F4.

With the rise of Large Language Models (LLMs), AI assistants' ability to utilize tools, especially through API calls, has advanced notably. This progress has necessitated more accurate evaluation methods. Many existing studies adopt static evaluation, where they assess AI assistants' API call based on pre-defined dialogue histories. However, such evaluation method can be misleading, as an AI assistant might fail in generating API calls from preceding human interaction in real cases. Instead of the resource-intensive method of direct human-machine interactions, we propose Automated Dynamic Evaluation (AutoDE) to assess an assistant's API call capability without human involvement. In our framework, we endeavor to closely mirror genuine human conversation patterns in human-machine interactions, using a LLM-based user agent, equipped with a user script to ensure human alignment. Experimental results highlight that AutoDE uncovers errors overlooked by static evaluations, aligning more closely with human assessment. Testing four AI assistants using our crafted benchmark, our method further mirrored human evaluation compared to conventional static evaluations.

Addressing the challenges related to data sparsity, cold-start problems, and diversity in recommendation systems is both crucial and demanding. Many current solutions leverage knowledge graphs to tackle these issues by combining both item-based and user-item collaborative signals. A common trend in these approaches focuses on improving ranking performance at the cost of escalating model complexity, reducing diversity, and complicating the task. It is essential to provide recommendations that are both personalized and diverse, rather than solely relying on achieving high rank-based performance, such as Click-through Rate, Recall, etc. In this paper, we propose a hybrid multi-task learning approach, training on user-item and item-item interactions. We apply item-based contrastive learning on descriptive text, sampling positive and negative pairs based on item metadata. Our approach allows the model to better understand the relationships between entities within the knowledge graph by utilizing semantic information from text. It leads to more accurate, relevant, and diverse user recommendations and a benefit that extends even to cold-start users who have few interactions with items. We perform extensive experiments on two widely used datasets to validate the effectiveness of our approach. Our findings demonstrate that jointly training user-item interactions and item-based signals using synopsis text is highly effective. Furthermore, our results provide evidence that item-based contrastive learning enhances the quality of entity embeddings, as indicated by metrics such as uniformity and alignment.

The majority of existing recommender systems rely on user ratings, which are limited by the lack of user collaboration and the sparsity problem. To address these issues, this study proposes a behavior-based recommender system that leverages customers' natural behaviors, such as browsing and clicking, on e-commerce platforms. The proposed recommendation system involves clustering active customers, determining neighborhoods, collecting similar users, calculating product reputation based on similar users, and recommending high-reputation products. To overcome the complexity of customer behaviors and traditional clustering methods, an unsupervised clustering approach based on product categories is developed to enhance the recommendation methodology. This study makes notable contributions in several aspects. Firstly, a groundbreaking behavior-based recommendation methodology is developed, incorporating customer behavior to generate accurate and tailored recommendations leading to improved customer satisfaction and engagement. Secondly, an original unsupervised clustering method, focusing on product categories, enables more precise clustering and facilitates accurate recommendations. Finally, an approach to determine neighborhoods for active customers within clusters is established, ensuring grouping of customers with similar behavioral patterns to enhance recommendation accuracy and relevance. The proposed recommendation methodology and clustering method contribute to improved recommendation performance, offering valuable insights for researchers and practitioners in the field of e-commerce recommendation systems. Additionally, the proposed method outperforms benchmark methods in experiments conducted using a behavior dataset from the well-known e-commerce site Alibaba.

Neural Network Language Models (NNLMs) for Virtual Assistants (VAs) are generally language-, region-, and in some cases, device-dependent, which increases the effort to scale and maintain them. Combining NNLMs for one or more of the categories is one way to improve scalability. In this work, we combine regional variants of English to build a ``World English'' NNLM for on-device VAs. In particular, we investigate the application of adapter bottlenecks to model dialect-specific characteristics in our existing production NNLMs {and enhance the multi-dialect baselines}. We find that adapter modules are more effective in modeling dialects than specializing entire sub-networks. Based on this insight and leveraging the design of our production models, we introduce a new architecture for World English NNLM that meets the accuracy, latency, and memory constraints of our single-dialect models.

Developing a unified model that can effectively harness heterogeneous resources and respond to a wide range of personalized needs has been a longstanding community aspiration. Our daily choices, especially in domains like fashion and retail, are substantially shaped by multi-modal data, such as pictures and textual descriptions. The vision and language modalities not only offer intuitive guidance but also cater to personalized user preferences. However, the predominant personalization approaches mainly focus on the ID or text-based recommendation problem, failing to comprehend the information spanning various tasks or modalities. In this paper, our goal is to establish a Unified paradigm for Multi-modal Personalization systems (UniMP), which effectively leverages multi-modal data while eliminating the complexities associated with task-and modality-specific customization. We argue that the advancements in foundational generative modeling have provided the flexibility and effectiveness necessary to achieve the objective. In light of this, we develop a generic and extensible personalization generative framework, that can handle a wide range of personalized needs including item recommendation, product search, preference prediction, explanation generation, and further userguided image generation. Our methodology enhances the capabilities of foundational language models for personalized tasks by seamlessly ingesting interleaved vision-language user history information, ensuring a more precise and customized experience for users. To train and evaluate the proposed multi-modal personalized tasks, we also introduce a novel and comprehensive benchmark covering a variety of user requirements. Our experiments on the real-world benchmark showcase the model's potential, outperforming competitive methods specialized for each task. With rapid growth, personalization systems have emerged as a key factor in meeting the user's expectations for tailored experiences that align with their unique needs and preferences. In today's digitally driven landscape, individuals engage with diverse data types, such as ratings, images, descriptions, and prices, especially in domains like fashion and retail (Kang et al., 2017; Hwangbo et al., 2018) where visuals and text are essential for decision-making. Given the profound influence of these multi-modal stimuli, there exists a pressing need for systems that can seamlessly integrate and harness these diverse data streams for improved personalization.

From London's cockney twang to Birmingham's Brummie accent, Britain is home to almost 40 regional dialects – each with its own unique words. Now, a study has revealed the quirky regional terms at risk of dying out. The South West's'ansum' and Yorkshire's'thoile' are just two of the terms that have declined in usage by up to 98 per cent over the last 100 years, according to researchers from SAS Northern Europe. 'As dialect is continuously changing the diversity and richness of our language, the quirks, idioms and phrases often associated with certain communities will continue to change over the next few years,' said Dr Iain Brown, Head of Data Science at SAS. So, how many of these regional words do you still use?

Smart home tech can increase the value of your home. Installing smart home devices is an easy way to protect your property from significant damage, save on home insurance, and modernize your home. Most insurance agents recommend devices like smoke, CO2, or water sensors to offer enhanced peace of mind and protection for homeowners and provide savings opportunities. Protecting a property from potentially extensive damage means homeowners are less likely to need to file a claim, which may result in lower insurance premiums. Smart home technology can also increase a home's value as appraisers factor smart home systems into appraisals and homes with intelligent technology may sell faster and command higher prices.