A5057643560- Topic Modeling
- Information Retrieval and Search Behavior
- Recommender Systems and Techniques
- Web Data Mining and Analysis
- Expert finding and Q&A systems
- Mobile Crowdsensing and Crowdsourcing
- Advanced Image and Video Retrieval Techniques
- Data Management and Algorithms
- Natural Language Processing Techniques
- Complex Network Analysis Techniques
- Text and Document Classification Technologies
- Advanced Text Analysis Techniques
- Algorithms and Data Compression
- Sentiment Analysis and Opinion Mining
- Advanced Graph Neural Networks
- Human Mobility and Location-Based Analysis
- Caching and Content Delivery
- Advanced Bandit Algorithms Research
- Misinformation and Its Impacts
- Spam and Phishing Detection
- Data Quality and Management
- Image Retrieval and Classification Techniques
- Biomedical Text Mining and Ontologies
- Semantic Web and Ontologies
- Multimodal Machine Learning Applications
University of Glasgow
2016-2025
Pratt Institute
2016
Dalhousie University
1993
When a Web user's underlying information need is not clearly specified from the initial query, an effective approach to diversify results retrieved for this query. In paper, we introduce novel probabilistic framework search result diversification, which explicitly accounts various aspects associated underspecified particular, document ranking by estimating how well given satisfies each uncovered aspect and extent different are satisfied as whole. We thoroughly evaluate our in context of...
Twitter is often considered to be a useful source of real-time news, potentially replacing newswire for this purpose. But true? In paper, we examine the extent which news reporting in and overlap whether reports faster than traditional providers. particular, analyse 77 days worth tweet articles with respect both manually identified major events larger volumes automatically events. Our results indicate that same as providers, addition long tail minor ignored by mainstream media. However,...
In an expert search task, the users' need is to identify people who have relevant expertise a topic of interest. An system predicts and ranks set candidate persons with respect query. this paper, we propose novel approach for predicting ranking We see problem experts as voting problem, which model by adapting eleven data fusion techniques.We investigate effectiveness associated techniques across range document weighting models, in context TREC 2005 Enterprise track. The evaluation results...
Venue recommendation systems aim to effectively rank a list of interesting venues users should visit based on their historical feedback (e.g. checkins). Such are increasingly deployed by Location-based Social Networks (LBSNs) such as Foursquare and Yelp enhance usefulness users. Recently, various RNN architectures have been proposed incorporate contextual information associated with the users' sequence checkins time day, location venues) capture dynamic preferences. However, these assume...
Miles Osborne, Sean Moran, Richard McCreadie, Alexander Von Lunen, Martin Sykora, Elizabeth Cano, Neil Ireson, Craig Macdonald, Iadh Ounis, Yulan He, Tom Jackson, Fabio Ciravegna, Ann O’Brien. Proceedings of 52nd Annual Meeting the Association for Computational Linguistics: System Demonstrations. 2014.
Ranking in information retrieval has been traditionally approached as a pursuit of relevant information, under the assumption that users' needs are unambiguously conveyed by their submitted queries.Nevertheless, an inherently limited representation more complex need, every query can arguably be considered ambiguous to some extent.In order tackle ambiguity, search result diversification approaches have recently proposed produce rankings aimed satisfy multiple possible underlying query.In this...
The users' historical interactions usually contain their interests and purchase habits based on which personalised recommendations can be made. However, such user are often sparse, leading to the well-known cold-start problem when a has no or very few interactions. In this paper, we propose new recommendation model, named Heterogeneous Graph Neural Recommender (HGNR), tackle while ensuring effective for all users. Our HGNR model learns users items' embeddings by using Convolutional Network...
PyTerrier is a Python-based retrieval framework for expressing simple and complex information (IR) pipelines in declarative manner. While making use of the long-established Terrier IR platform basic text indexing retrieval, its salient utility comes from expressive Python operators, which allow individual operations to be pipelined combined different flexible manners as requested by search application. Each operation applies transformation upon dataframe, while operators are defined with...
Recently micro-videos have become more popular in social media platforms such as TikTok and Instagram. Engagements these are facilitated by multi-modal recommendation systems. Indeed, multimedia content can involve diverse modalities, often represented visual, acoustic, textual features to the recommender model. Existing works micro-video tend unify channels, thereby treating each modality with equal importance. However, we argue that approaches not sufficient encode item representations...
The Twitter real-time information network is the subject of research for retrieval tasks such as search. However, so far, reproducible experimentation on data has been impeded by restrictions imposed terms service. In this paper, we detail a new methodology legally building and distributing corpora, developed through collaboration between Text REtrieval Conference (TREC) Twitter. particular, how first publicly available corpus - referred to Tweets2011 was distributed via lists tweet...
Search result diversification has gained momentum as a way to tackle ambiguous queries. An effective approach this problem is explicitly model the possible aspects underlying query, in order maximise estimated relevance of retrieved documents with respect different aspects. However, such themselves may represent information needs rather distinct intents (e.g., informational or navigational). Hence, diverse ranking could benefit from applying intent-aware retrieval models when estimating In...
Venue recommendation is an important application for Location-Based Social Networks (LBSNs), such as Yelp, and has been extensively studied in recent years. Matrix Factorisation (MF) a popular Collaborative Filtering (CF) technique that can suggest relevant venues to users based on assumption similar are likely visit venues. In years, deep neural networks have successfully applied tasks speech recognition, computer vision natural language processing. Building upon this momentum, various...
Social media streams, such as Twitter, have shown themselves to be useful sources of real-time information about what is happening in the world. Automatic detection and tracking events identified these streams a variety real-world applications, e.g. identifying automatically reporting road accidents for emergency services. However, useful, need within stream with very low latency. This challenging due high volume posts social streams. In this paper, we propose novel event approach that can...