Mariam Daoud

Bio: Mariam Daoud is an academic researcher from York University. The author has contributed to research in topics: Ranking (information retrieval) & Query optimization. The author has an hindex of 9, co-authored 21 publications receiving 349 citations. Previous affiliations of Mariam Daoud include Seneca College.

Evaluation of contextual information retrieval effectiveness: overview of issues and research

Abstract: The increasing prominence of information arising from a wide range of sources delivered over electronic media has made traditional information retrieval systems less effective. Indeed, users are overwhelmed by the information delivered by such systems in response to their queries, particularly when the latter are ambiguous. In order to tackle this problem, the state-of-the-art reveals that there is a growing interest towards contextual information retrieval which relies on various sources of evidence issued from the user’s search background and environment like interests, preferences, time and location, in order to improve the retrieval accuracy. Contextual information retrieval systems are based on different definitions of the core concept of user’s context, various user’s context modeling approaches and several techniques of document relevance measurement, but all share the goal of providing the most useful information to the users in accordance with their context. However, the evaluation methodologies conceived in the past several years for traditional information retrieval and widely used in the evaluation campaigns have been challenged by the consideration of user’s context in the information retrieval process. Thus, we recognize that a critical review of existing evaluation methodologies in contextual information retrieval area is needed in order to design and develop standard evaluation frameworks. We present in this paper a comprehensive survey of contextual information retrieval evaluation methodologies and provide insights into how and why they are appropriate to measure the retrieval effectiveness. We also highlight some of the research challenges ahead that would constitute substantive research area for future research.

Learning user interests for a session-based personalized search

Abstract: It is now widely assumed in personalized information retrieval (IR) area that user interests can provide substantial clues for document relevance estimation. User interests reflect generally the user background and topics of interests. However most of the proposed personalized retrieval models and strategies do not distinguish between short term and long term user interests and make use of the whole search history to improve the search accuracy. In this paper, we study how to learn long term user interests by aggregating concept-based short term ones identified within related search activities. For this purpose, we tackle the problem of session boundary recognition using context-sensitive similarity measures that are able to gauge the changes in the user interest topics with regard to reference ontology. Finally, the search personalization is achieved by re-ranking the search results for a given query using the short term user interest. Our experimental evaluation is carried out using TREC collection and shows that personalization brings significant improvements in retrieval effectiveness. Moreover, we observe that our context-sensitive session boundary recognition method can, to some extent, find a semantic correlation between the query and the user context across the search sessions.

Evaluation in Contextual Information Retrieval: Foundations and Recent Advances within the Challenges of Context Dynamicity and Data Privacy

Abstract: Context such as the user’s search history, demographics, devices, and surroundings, has become prevalent in various domains of information seeking and retrieval such as mobile search, task-based search, and social search While evaluation is central and has a long history in information retrieval, it faces the big challenge of designing an appropriate methodology that embeds the context into evaluation settings In this article, we present a unified summary of a wide range of main and recent progress in contextual information retrieval evaluation that leverages diverse context dimensions and uses different principles, methodologies, and levels of measurements More specifically, this survey article aims to fill two main gaps in the literature: First, it provides a critical summary and comparison of existing contextual information retrieval evaluation methodologies and metrics according to a simple stratification model; second, it points out the impact of context dynamicity and data privacy on the evaluation design Finally, we recommend promising research directions for future investigations

Exploiting semantics for improving clinical information retrieval

Abstract: Clinical information retrieval (IR) presents several challenges including terminology mismatch and granularity mismatch. One of the main objectives in clinical IR is to fill the semantic gap among the queries and documents and go beyond keywords matching. To address these issues, in this paper we attempt to use semantic information to improve the performance of clinical IR systems by representing queries in an expressive and meaningful context. To model a query context initially we model and develop query domain ontology. The query domain ontology represents concepts closely related with query concepts. Query context represents concepts extracted from query domain ontology and weighted according to their semantic relatedness to query concept(s). The query context is then exploited in query expansion and patients records re-ranking for improving clinical retrieval performance. We evaluate our approach on the TREC Medical Records dataset. Results show that our proposed approach significantly improves the retrieval performance compare to classic keyword-based IR model.

Learning implicit user interests using ontology and search history for personalization

Abstract: The key for providing a robust context for personalized information retrieval is to build a library which gathers the long term and the short term user's interests and then using it in the retrieval process in order to deliver results that better meet the user's information needs. In this paper, we present an enhanced approach for learning a semantic representation of the underlying user's interests using the search history and a predefined ontology. The basic idea is to learn the user's interests by collecting evidence from his search history and represent them conceptually using the concept hierarchy of the ontology. We also involve a dynamic method which tracks changes of the short term user's interests using a correlation metric measure in order to learn and maintain the user's interests.

Data preprocessing techniques for classification without discrimination

Abstract: Recently, the following Discrimination-Aware Classification Problem was introduced: Suppose we are given training data that exhibit unlawful discrimination; e.g., toward sensitive attributes such as gender or ethnicity. The task is to learn a classifier that optimizes accuracy, but does not have this discrimination in its predictions on test data. This problem is relevant in many settings, such as when the data are generated by a biased decision process or when the sensitive attribute serves as a proxy for unobserved features. In this paper, we concentrate on the case with only one binary sensitive attribute and a two-class classification problem. We first study the theoretically optimal trade-off between accuracy and non-discrimination for pure classifiers. Then, we look at algorithmic solutions that preprocess the data to remove discrimination before a classifier is learned. We survey and extend our existing data preprocessing techniques, being suppression of the sensitive attribute, massaging the dataset by changing class labels, and reweighing or resampling the data to remove discrimination without relabeling instances. These preprocessing techniques have been implemented in a modified version of Weka and we present the results of experiments on real-life data.

Personalizing web search using long term browsing history

Abstract: Personalizing web search results has long been recognized as an avenue to greatly improve the search experience. We present a personalization approach that builds a user interest profile using users' complete browsing behavior, then uses this model to rerank web results. We show that using a combination of content and previously visited websites provides effective personalization. We extend previous work by proposing a number of techniques for filtering previously viewed content that greatly improve the user model used for personalization. Our approaches are compared to previous work in offline experiments and are evaluated against unpersonalized web search in large scale online tests. Large improvements are found in both cases.

Near-Eye Display and Tracking Technologies for Virtual and Augmented Reality

Abstract: Virtual and augmented reality (VR/AR) are expected to revolutionise entertainment, healthcare, communication and the manufacturing industries among many others. Near-eye displays are an enabling vessel for VR/AR applications, which have to tackle many challenges related to ergonomics, comfort, visual quality and natural interaction. These challenges are related to the core elements of these near-eye display hardware and tracking technologies. In this state-of-the-art report, we investigate the background theory of perception and vision as well as the latest advancements in display engineering and tracking technologies. We begin our discussion by describing the basics of light and image formation. Later, we recount principles of visual perception by relating to the human visual system. We provide two structured overviews on state-of-the-art near-eye display and tracking technologies involved in such near-eye displays. We conclude by outlining unresolved research questions to inspire the next generation of researchers.

Evaluation of contextual information retrieval effectiveness: overview of issues and research

Abstract: The increasing prominence of information arising from a wide range of sources delivered over electronic media has made traditional information retrieval systems less effective. Indeed, users are overwhelmed by the information delivered by such systems in response to their queries, particularly when the latter are ambiguous. In order to tackle this problem, the state-of-the-art reveals that there is a growing interest towards contextual information retrieval which relies on various sources of evidence issued from the user’s search background and environment like interests, preferences, time and location, in order to improve the retrieval accuracy. Contextual information retrieval systems are based on different definitions of the core concept of user’s context, various user’s context modeling approaches and several techniques of document relevance measurement, but all share the goal of providing the most useful information to the users in accordance with their context. However, the evaluation methodologies conceived in the past several years for traditional information retrieval and widely used in the evaluation campaigns have been challenged by the consideration of user’s context in the information retrieval process. Thus, we recognize that a critical review of existing evaluation methodologies in contextual information retrieval area is needed in order to design and develop standard evaluation frameworks. We present in this paper a comprehensive survey of contextual information retrieval evaluation methodologies and provide insights into how and why they are appropriate to measure the retrieval effectiveness. We also highlight some of the research challenges ahead that would constitute substantive research area for future research.