Zaigham Faraz Siddiqui

Discovering global and local bursts in a stream of news

Reports on major events like hurricanes and earthquakes, and major topics like the financial crisis or the Egyptian revolution appear in Internet news and become (ir)regularly updated, as new insights are acquired. Tracking emerging subtopics in a major or even local event is important for the news readers but challenging for the operator: subtopics may emerge gradually or in a bursty way; they may be of some importance inside the event, but too rare to be visible inside the whole stream of news. In this study, we propose a text stream clustering method that detects, tracks and updates large and small bursts of news in a two-level topic hierarchy. We report on our first results on a stream of news from February to April 2011.

Online clustering of high-dimensional trajectories under concept drift

Historical transaction data are collected in many applications, e.g., patient histories recorded by physicians and customer transactions collected by companies. An important question is the learning of models upon the primary objects (patients, customers) rather than the transactions, especially when these models are subjected to drift. We address this problem by combining advances of online clustering on multivariate data with the trajectory mining paradigm. We model the measurements of each individual primary object (e.g. its transactions), taken at irregular time intervals, as a trajectory in a high-dimensional feature space. Then, we cluster individuals with similar trajectories to identify sub-populations that evolve similarly, e.g. groups of customers that evolve similarly or groups of employees that have similar careers. We assume that the multivariate trajectories are generated by drifting Gaussian Mixture Models. We study (i) an EM-based approach that clusters these trajectories incrementally as a reference method that has access to all the data for learning, and propose (ii) an online algorithm based on a Kalman filter that efficiently tracks the trajectories of Gaussian clusters. We show that while both methods approximate the reference well, the algorithm based on a Kalman filter is faster by one order of magnitude compared to the EM-based approach.

Where are we going? predicting the evolution of individuals

When searching for patterns on data streams, we come across perennial (dynamic) objects that evolve over time. These objects are encountered repeatedly and each time with different definition and values. Examples are (a) companies registered at stock exchange and reporting their progress at the end of each year, and (b) students whose performance is evaluated at the end of each semester. On such data, domain experts also pose questions on how the individual objects will evolve: would it be beneficial to invest in a given company, given both the companys individual performance thus far and the drift experienced in the model? Or, how will a given student perform next year, given the performance variations observed thus far? While there is much research on how models evolve/change over time [Ntoutsi et al., 2011a], little is done to predict the change of individual objects when the states are not known a priori. In this work, we propose a framework that learns the clusters to which the objects belong at each moment, uses them as ad hoc states in a state-transition graph, and then learns a mixture model of Markov Chains, which predicts the next most likely state/cluster per object. We report on our evaluation on synthetic and real datasets.

Learning Relational User Profiles and Recommending Items as Their Preferences Change

Over the last decade a vast number of businesses have developed online e-shops in the web. These online stores are supported by sophisticated systems that manage the products and record the activity of customers. There exist many research works that strive to answer the question “what items are the customers going to like” given their historical profiles. However, most of these works do not take into account the time dimension and cannot respond efficiently when data are huge. In this paper, we study the problem of recommendations in the context of multi-relational stream mining. Our algorithm “xStreams” first separates customers based on their historical data into clusters. It then employs collaborative filtering (CF) to recommend new items to the customers based on their group similarity. To evaluate the working of xStreams, we use a multi-relational data generator for streams. We evaluate xStreams on real and synthetic datasets.

Classification rule mining for a stream of perennial objects

We study classification over a slow stream of complex objects like customers or students. The learning task must take into account that an objects label is influenced by incoming data from adjoint, fast streams of transactions, e.g. customer purchases or student exams, and that this label may even change over time. This task involves combining the streams, and exploiting associations between the target label and attribute values in the fast streams. We propose a method for the discovery of classification rules over such a confederation of streams, and we use it to enhance a decision tree classifier. We show that the new approach has competitive predictive power while building much smaller decision trees than the original classifier.

Are Some Brain Injury Patients Improving More Than Others

Predicting the evolution of individuals is a rather new mining task with applications in medicine. Medical researchers are interested in the progress of a disease and in the evolution of individuals subjected to treatment. We investigate the evolution of patients on the basis of medical tests before and during treatment after brain trauma: we want to understand how similar patients can become to healthy participants. We face two challenges. First, we have less information on healthy participants than on the patients. Second, the values of the medical tests for patients, even after treatment started, remain well-separated from those of healthy people; this is typical for neurodegenerative diseases, but also for further brain impairments. Our approach encompasses methods for modelling patient evolution and for predicting the health improvement of different patient subpopulations, dealing with the above challenges. We test our approach on a cohort of patients treated after brain trauma and a corresponding cohort of controls.

Predicting the post-treatment recovery of patients suffering from traumatic brain injury (TBI)

Predicting the evolution of individuals is a rather new mining task with applications in medicine. Medical researchers are interested in the progression of a disease and/or how do patients evolve or recover when they are subjected to some treatment. In this study, we investigate the problem of patients’ evolution on the basis of medical tests before and after treatment after brain trauma: we want to understand to what extend a patient can become similar to a healthy participant. We face two challenges. First, we have less information on healthy participants than on the patients. Second, the values of the medical tests for patients, even after treatment started, remain well-separated from those of healthy people; this is typical for neurodegenerative diseases, but also for further brain impairments. Our approach encompasses methods for modelling patient evolution and for predicting the health improvement of different patients’ subpopulations, i.e. prediction of label if they recovered or not. We test our approach on a cohort of patients treated after brain trauma and a corresponding cohort of controls.