Personalized recommendation systems often drive users towards more extreme content, exacerbating opinion polarization. While content-aware moderation has been proposed to mitigate these effects, such approaches risk curtailing the freedom of speech and information. To address this concern, we propose and explore the feasibility of content-agnostic moderation as an alternative approach for reducing polarization. Content-agnostic moderation does not rely on the actual content being moderated, arguably making it less prone to forms of censorship. We establish theoretically that content-agnostic moderation cannot be guaranteed to work in a fully generic setting. However, we show that it can often be effectively achieved in practice with plausible assumptions. We introduce two novel content-agnostic moderation methods that modify recommendations from the content recommender to disperse user-item co-clusters without relying on content features.

Dwell time (DT) is a critical post-click metric for evaluating user preference in recommender systems, complementing the traditional click-through rate (CTR). Although multi-task learning is widely adopted to jointly optimize DT and CTR, we observe that multi-task models systematically collapse their DT predictions to the shortest and longest bins, under-predicting the moderate durations. We attribute this moderate-duration bin under-representation to over-reliance on the CTR-DT spurious correlation, and propose ORCA to address it with causal-decoupling. Specifically, ORCA explicitly models and subtracts CTR's negative transfer while preserving its positive transfer. We further introduce (i) feature-level counterfactual intervention, and (ii) a task-interaction module with instance inverse-weighting, weakening CTR-mediated effect and restoring direct DT semantics. ORCA is model-agnostic and easy to deploy. Experiments show an average 10.6% lift in DT metrics without harming CTR. Code is available at https://github.com/Chrissie-Law/ORCA-Mitigating-Over-Reliance-for-Multi-Task-Dwell-Time-Prediction-with-Causal-Decoupling.

Abstract Temporal networks allow representing connections between objects while incorporating the temporal dimension. While static network models can capture unchanging topological regularities, they often fail to model the effects associated with the causal generative process of the network that occurs in time. Hence, exploiting the temporal aspect of networks has been the focus of many recent studies. In this context, we propose a new framework for generative models of continuous-time temporal networks. We assume that the activation of the edges in a temporal network is driven by a specified temporal point process. This approach allows to directly model the waiting time between events while incorporating time-varying history-based features as covariates in the predictions. Coupled with a thinning algorithm designed for the simulation of point processes, SimHawNet enables simulation of the evolution of temporal networks in continuous time. Finally, we introduce a comprehensive evaluation framework to assess the performance of such an approach, in which we demonstrate that SimHawNet successfully simulates the evolution of networks with very different generative processes and achieves performance comparable to the state of the art, while being significantly faster.

Recommendation in settings such as e-recruitment and online dating involves distributing limited opportunities, which differs from recommending practically unlimited goods such as in e-commerce or music recommendation. This setting calls for novel approaches to quantify and enforce fairness. Indeed, typical recommender systems recommend each user their top relevant items, such that desirable items may be recommended simultaneously to more and to less qualified individuals. This is arguably unfair to the latter. Indeed, when they pursue such a desirable recommendation (e.g., by applying for a job), they are unlikely to be successful. To quantify fairness in such settings, we introduce inferiority : a novel (un)fairness measure that quantifies the competitive disadvantage of a user for their recommended items. Inferiority is complementary to envy : a previously-proposed fairness notion that quantifies the extent to which a user prefers other users’ recommendations over their own. We propose to use both inferiority and envy in combination with an accuracy-related measure called utility : the aggregated relevancy scores of the recommended items. Unfortunately, none of these three measures are differentiable, making it hard to optimize them, and restricting their immediate use to evaluation only. To remedy this, we reformulate them in the context of a probabilistic interpretation of recommender systems, resulting in differentiable versions. We show how these loss functions can be combined in a multi-objective optimization problem that we call FEIR (Fairness through Envy and Inferiority Reduction), used as a post-processing of the scores from any standard recommender system. Experiments on synthetic and real-world data show that the proposed approach effectively improves the trade-offs between inferiority, envy and utility, compared to the naive recommendation and the state-of-the-art method for the related problem of congestion alleviation in job recommendation. We discuss and enhance the practical impact of our findings on a wide range of real-world recommendation scenarios, and we offer implementations of visualization tools to render the envy and inferiority metrics more accessible.

Real-world data can often be represented as a heterogeneous network relating nodes of different types. E.g., a job market can be represented as a job seeker-skill-vacancy network. It can be relevant to consider theimbalancebetween nodes of different types, in terms of whether they are similarly connected in the network. For example, it is desirable that job seekers and vacancies are mixed well. If they are not, then there is imbalance. We propose to quantify the imbalancebetween two sets of nodesin a network as the Earth Mover's Distance between the sets. Given this quantification, we introduceGREASE(Graph imbalance REduction by Adding Sets of Edges), a method that selects a fixed number of unconnected node-pairs, which—if links were added between them—aims to maximally reduce the imbalance. In the job market network,GREASEcan be used to select skills that job seekers do not yet have, but could strive to acquire, to reduce the imbalance between job seekers and vacancies.GREASEmay also be used in other applications, such as reducing controversy between opposing sides on a polarizing topic. We evaluatedGREASEon several datasets and find thatGREASEoutperforms baselines in reducing network imbalance.

Recommender systems may suffer from congestion, meaning that there is an unequal distribution of the items in how often they are recommended. Some items may be recommended much more than others. Recommenders are increasingly used in domains where items have limited availability, such as the job market, where congestion is especially problematic: Recommending a vacancy—for which typically only one person will be hired—to a large number of job seekers may lead to frustration for job seekers, as they may be applying for jobs where they are not hired. This may also leave vacancies unfilled and result in job market inefficiency.

Dimensionality reduction and manifold learning methods such as t-distributed Stochastic Neighbor Embedding (t-SNE) are frequently used to map high-dimensional data into a two-dimensional space to visualize and explore that data. Going beyond the specifics of t-SNE, there are two substantial limitations of any such approach: (1) not all information can be captured in a single two-dimensional embedding, and (2) to well-informed users, the salient structure of such an embedding is often already known, preventing that any real new insights can be obtained. Currently, it is not known how to extract the remaining information in a similarly effective manner. We introduce conditional t-SNE (ct-SNE), a generalization of t-SNE that discounts prior information in the form of labels. This enables obtaining more informative and more relevant embeddings. To achieve this, we propose a conditioned version of the t-SNE objective, obtaining an elegant method with a single integrated objective. We show how to efficiently optimize the objective and study the effects of the extra parameter that ct-SNE has over t-SNE. Qualitative and quantitative empirical results on synthetic and real data show ct-SNE is scalable, effective, and achieves its goal: it allows complementary structure to be captured in the embedding and provides new insights into data.

Abstract The connectivity structure of graphs is typically related to the attributes of the vertices. In social networks for example, the probability of a friendship between any pair of people depends on a range of attributes, such as their age, residence location, workplace, and hobbies. The high-level structure of a graph can thus possibly be described well by means of patterns of the form ‘the subgroup of all individuals with certain properties X are often (or rarely) friends with individuals in another subgroup defined by properties Y’, ideally relative to their expected connectivity. Such rules present potentially actionable and generalizable insight into the graph. Prior work has already considered the search for dense subgraphs (‘communities’) with homogeneous attributes. The first contribution in this paper is to generalize this type of pattern to densities between apair of subgroups, as well as betweenall pairs from a set of subgroups that partition the vertices. Second, we develop a novel information-theoretic approach for quantifying the subjective interestingness of such patterns, by contrasting them with prior information an analyst may have about the graph’s connectivity. We demonstrate empirically that in the special case of dense subgraphs, this approach yields results that are superior to the state-of-the-art. Finally, we propose algorithms for efficiently finding interesting patterns of these different types.

Real-world data often presents itself in the form of a network. Examples include social networks, citation networks, biological networks, and knowledge graphs. In their simplest form, networks represent real-life entities (e.g. people, papers, proteins, concepts) as nodes, and describe them in terms of their relations with other entities by means of edges between these nodes. This can be valuable for a range of purposes from the study of information diffusion to bibliographic analysis, bioinformatics research, and question-answering. The quality of networks is often problematic though, affecting downstream tasks. This paper focuses on the common problem where a node in the network in fact corresponds to multiple real-life entities. In particular, we introduce FONDUE, an algorithm based on network embedding for node disambiguation. Given a network, FONDUE identifies nodes that correspond to multiple entities, for subsequent splitting. Extensive experiments on fourteen benchmark datasets demonstrate that FONDUE is substantially and uniformly more accurate for ambiguous node identification compared to the existing state-of-the-art with a lower computational cost, while less optimal for determining the best way to split ambiguous nodes.

The connectivity structure of graphs is typically related to the attributes of the nodes. In social networks for example, the probability of a friendship between two people depends on their attributes, such as their age, address, and hobbies. The connectivity of a graph can thus possibly be understood in terms of patterns of the form ‘the subgroup of individuals with properties X are often (or rarely) friends with individuals in another subgroup with properties Y'. Such rules present potentially actionable and generalizable insights into the graph. We present a method that finds pairs of node subgroups between which the edge density is interestingly high or low, using an information-theoretic definition of interestingness. This interestingness is quantified subjectively, to contrast with prior information an analyst may have about the graph. This view immediately enables iterative mining of such patterns. Our work generalizes prior work on dense subgraph mining (i.e. subgraphs induced by a single subgroup). Moreover, not only is the proposed method more general, we also demonstrate considerable practical advantages for the single subgroup special case.

Visual exploration of high-dimensional real-valued datasets is a fundamental task in exploratory data analysis (EDA). Existing methods use predefined criteria to choose the representation of data. There is a lack of methods that (i) elicit from the user what she has learned from the data and (ii) show patterns that she does not know yet. We construct a theoretical model where identified patterns can be input as knowledge to the system. The knowledge syntax here is intuitive, such as "this set of points forms a cluster", and requires no knowledge of maths. This background knowledge is used to find a Maximum Entropy distribution of the data, after which the system provides the user data projections in which the data and the Maximum Entropy distribution differ the most, hence showing the user aspects of the data that are maximally informative given the user's current knowledge. We provide an open source EDA system with tailored interactive visualizations to demonstrate these concepts. We study the performance of the system and present use cases on both synthetic and real data. We find that the model and the prototype system allow the user to learn information efficiently from various data sources and the system works sufficiently fast in practice. We conclude that the information theoretic approach to exploratory data analysis where patterns observed by a user are formalized as constraints provides a principled, intuitive, and efficient basis for constructing an EDA system.

Data visualization and iterative/interactive data mining are growing rapidly in attention, both in research as well as in industry. However, while there are a plethora of advanced data mining methods and lots of works in the field of visualization, integrated methods that combine advanced visualization and/or interaction with data mining techniques in a principled way are rare. We present a framework based on constrained randomization which lets users explore high-dimensional data via `subjectively informative' two-dimensional data visualizations. The user is presented with `interesting' projections, allowing users to express their observations using visual interactions that update a background model representing the user's belief state. This background model is then considered by a projection-finding algorithm employing data randomization to compute a new `interesting' projection. By providing users with information that contrasts with the background model, we maximize the chance that the user encounters striking new information present in the data. This process can be iterated until the user runs out of time or until the difference between the randomized and the real data is insignificant. We present two case studies, one controlled study on synthetic data and another on census data, using the proof-of-concept tool SIDE that demonstrates the presented framework.

Deriving insights from high-dimensional data is one of the core problems in data mining. The difficulty mainly stems from the large number of variable combinations to potentially consider. Hence, an obvious question is whether we can automate the search for interesting patterns. Here, we consider the setting where a user wants to learn as efficiently as possible about real-valued attributes. We introduce a method to find subgroups in the data that are maximally informative (in the Information Theoretic sense) with respect to one or more real-valued target attributes. The succinct subgroup descriptions are in terms of arbitrarily-typed description attributes. The approach is based on the Subjective Interestingness framework FORSIED to use prior knowledge when mining most informative patterns.

The information in high-dimensional datasets is often too complex for human users to perceive directly. Hence, it may be helpful to use dimensionality reduction methods to construct lower dimensional representations that can be visualized. The natural question that arises is how do we construct a most informative low dimensional representation? We study this question from an information-theoretic perspective and introduce a new method for linear dimensionality reduction. The obtained model that quantifies the informativeness also allows us to flexibly account for prior knowledge a user may have about the data. This enables us to provide representations that are subjectively interesting . We title the method Subjectively Interesting Component Analysis (SICA) and expect it is mainly useful for iterative data mining. SICA is based on a model of a user’s belief state about the data. This belief state is used to search for surprising views. The initial state is chosen by the user (it may be empty up to the data format) and is updated automatically as the analysis progresses. We study several types of prior beliefs: if a user only knows the scale of the data, SICA yields the same cost function as Principal Component Analysis (PCA), while if a user expects the data to have outliers, we obtain a variant that we term t -PCA. Finally, scientifically more interesting variants are obtained when a user has more complicated beliefs, such as knowledge about similarities between data points. The experiments suggest that SICA enables users to find subjectively more interesting representations.

Methods that find insightful low-dimensional projections are essential to effectively explore high-dimensional data. Principal Component Analysis is used pervasively to find low-dimensional projections, not only because it is straightforward to use, but it is also often effective, because the variance in data is often dominated by relevant structure. However, even if the projections highlight real structure in the data, not all structure is interesting to every user. If a user is already aware of, or not interested in the dominant structure, Principal Component Analysis is less effective for finding interesting components. We introduce a new method called Subjectively Interesting Component Analysis (SICA), designed to find data projections that are subjectively interesting, i.e, projections that truly surprise the end-user. It is rooted in information theory and employs an explicit model of a user's prior expectations about the data. The corresponding optimization problem is a simple eigenvalue problem, and the result is a trade-off between explained variance and novelty. We present five case studies on synthetic data, images, time-series, and spatial data, to illustrate how SICA enables users to find (subjectively) interesting projections.

Local pattern mining methods are fragmented along two dimensions: the pattern syntax, and the data types on which they are applicable. Pattern syntaxes considered in the literature include subgroups, n-sets, itemsets, and many more; common data types include binary, categorical, and real-valued. Recent research on pattern mining in relational databases has shown how the aforementioned pattern syntaxes can be unified in a single framework. However, a unified understanding of how to deal with various data types is lacking, certainly for more complexly structured types such as time of day (which is circular), geographical location, terms from a taxonomy, etc. In this paper, we introduce a generic approach for mining interesting local patterns in (relational) data involving such structured data types as attributes. Importantly, we show how this can be done in a generic manner, by modelling the structure within a set of attribute values as a partial order. We then derive a measure of subjective interestingness of such patterns using Information Theory, and propose an algorithm for effectively enumerating all patterns of this syntax. Through empirical evaluation, we found that (a) the new interestingness derivation is relevant and cannot be approximated using existing tools, (b) the new tool, P-N-RMiner, finds patterns that are substantially more informative, and (c) the new enumeration algorithm is considerably faster.