Computer Science

Our work aimed at experimentally assessing the benefits of model ensembling within the context of neural methods for passage reranking. Starting from relatively standard neural models, we use a previous technique named Fast Geometric Ensembling to generate multiple model instances from particular training schedules, then focusing or attention on different types of approaches for combining the results from the multiple model instances (e.g., averaging the ranking scores, using fusion methods from the IR literature, or using supervised learning-to-rank). Tests with the MS-MARCO dataset show that model ensembling can indeed benefit the ranking quality, particularly with supervised learning-to-rank although also with unsupervised rank aggregation.

This article analyses and evaluates FDD\b{eta}, a supervised term-weighting scheme that can be applied for query-term selection in topic-based retrieval. FDD\b{eta} weights terms based on two factors representing the descriptive and discriminating power of the terms with respect to the given topic. It then combines these two factor through the use of an adjustable parameter that allows to favor different aspects of retrieval, such as precision, recall or a balance between both. The article makes the following contributions: (1) it presents an extensive analysis of the behavior of FDD\b{eta} as a function of its adjustable parameter; (2) it compares FDD\b{eta} against eighteen traditional and state-of-the-art weighting scheme; (3) it evaluates the performance of disjunctive queries built by combining terms selected using the analyzed methods; (4) it introduces a new public data set with news labeled as relevant or irrelevant to the economic domain. The analysis and evaluations are performed on three data sets: two well-known text data sets, namely 20 Newsgroups and Reuters-21578, and the newly released data set. It is possible to conclude that despite its simplicity, FDD\b{eta} is competitive with state-of-the-art methods and has the important advantage of offering flexibility at the moment of adapting to specific task goals. The results also demonstrate that FDD\b{eta} offers a useful mechanism to explore different approaches to build complex queries.

Assessors make preference judgments faster and more consistently than graded judgments. Preference judgments can also recognize distinctions between items that appear equivalent under graded judgments. Unfortunately, preference judgments can require more than linear effort to fully order a pool of items, and evaluation measures for preference judgments are not as well established as those for graded judgments, such as NDCG. In this paper, we explore the assessment process for partial preference judgments, with the aim of identifying and ordering the top items in the pool, rather than fully ordering the entire pool. To measure the performance of a ranker, we compare its output to this preferred ordering by applying a rank similarity measure.We demonstrate the practical feasibility of this approach by crowdsourcing partial preferences for the TREC 2019 Conversational Assistance Track, replacing NDCG with a new measure named "compatibility". This new measure has its most striking impact when comparing modern neural rankers, where it is able to recognize significant improvements in quality that would otherwise be missed by NDCG.

With advances in the field of machine learning, precisely algorithms for recommendation systems, robot assistants are envisioned to become more present in the hospitality industry. Additionally, the COVID-19 pandemic has also highlighted the need to have more service robots in our everyday lives, to minimise the risk of human to-human transmission. One such example would be coffee shops, which have become intrinsic to our everyday lives. However, serving an excellent cup of coffee is not a trivial feat as a coffee blend typically comprises rich aromas, indulgent and unique flavours and a lingering aftertaste. Our work addresses this by proposing a computational model which recommends optimal coffee beans resulting from the user's preferences. Specifically, given a set of coffee bean properties (objective features), we apply different supervised learning techniques to predict coffee qualities (subjective features). We then consider an unsupervised learning method to analyse the relationship between coffee beans in the subjective feature space. Evaluated on a real coffee beans dataset based on digitised reviews, our results illustrate that the proposed computational model gives up to 92.7 percent recommendation accuracy for coffee beans prediction. From this, we propose how this computational model can be deployed on a service robot to reliably predict customers' coffee bean preferences, starting from the user inputting their coffee preferences to the robot recommending the coffee beans that best meet the user's likings.

Users prefer diverse recommendations over homogeneous ones. However, most previous work on Sequential Recommenders does not consider diversity, and strives for maximum accuracy, resulting in homogeneous recommendations. In this paper, we consider both accuracy and diversity by presenting an Attribute-aware Diversifying Sequential Recommender (ADSR). Specifically, ADSR utilizes available attribute information when modeling a user's sequential behavior to simultaneously learn the user's most likely item to interact with, and their preference of attributes. Then, ADSR diversifies the recommended items based on the predicted preference for certain attributes. Experiments on two benchmark datasets demonstrate that ADSR can effectively provide diverse recommendations while maintaining accuracy.

Machine Learning models are increasingly being adopted in many applications. The quality of these models critically depends on the input data on which they are trained, and by augmenting their input data with external data, we have the opportunity to create better models. However, the massive number of datasets available on the Web makes it challenging to find data suitable for augmentation. In this demo, we present our ongoing efforts to develop a dataset search engine tailored for data augmentation. Our prototype, named Auctus, automatically discovers datasets on the Web and, different from existing dataset search engines, infers consistent metadata for indexing and supports join and union search queries. Auctus is already being used in a real deployment environment to improve the performance of ML models. The demonstration will include various real-world data augmentation examples and visitors will be able to interact with the system.

Over the past years, there has been many efforts to curate and increase the added value of the raw data. Data curation has been defined as activities and processes an analyst undertakes to transform the raw data into contextualized data and knowledge. Data curation enables decision-makers and data analyst to extract value and derive insight from the raw data. However, to curate the raw data, an analyst needs to carry out various curation tasks including, extraction linking, classification, and indexing, which are error-prone, tedious and challenging. Besides, deriving insight require analysts to spend a long period of time to scan and analyze the curation environments. This problem is exacerbated when the curation environment is large, and the analyst needs to curate a varied and comprehensive list of data. To address these challenges, in this dissertation, we present techniques, algorithms and systems for augmenting analysts in curation tasks. We propose: ~(1) a feature-based and automated technique for curating the raw data. ~(2) We propose an autonomic approach for adapting data curation rules. ~(3) We provide a solution to augment users in formulating their preferences while curating data in large scale information spaces. ~(4) We implement a set of APIs for automating the basic curation tasks, including Named Entity extraction, POS tags, classification, and etc.

We introduce Auto-Surprise, an Automated Recommender System library. Auto-Surprise is an extension of the Surprise recommender system library and eases the algorithm selection and configuration process. Compared to out-of-the-box Surprise library, Auto-Surprise performs better when evaluated with MovieLens, Book Crossing and Jester Datasets. It may also result in the selection of an algorithm with significantly lower runtime. Compared to Surprise's grid search, Auto-Surprise performs equally well or slightly better in terms of RMSE, and is notably faster in finding the optimum hyperparameters.

Realistic recommender systems are often required to adapt to ever-changing data and tasks or to explore different models systematically. To address the need, we present AutoRec, an open-source automated machine learning (AutoML) platform extended from the TensorFlow ecosystem and, to our knowledge, the first framework to leverage AutoML for model search and hyperparameter tuning in deep recommendation models. AutoRec also supports a highly flexible pipeline that accommodates both sparse and dense inputs, rating prediction and click-through rate (CTR) prediction tasks, and an array of recommendation models. Lastly, AutoRec provides a simple, user-friendly API. Experiments conducted on the benchmark datasets reveal AutoRec is reliable and can identify models which resemble the best model without prior knowledge.

Sentic computing relies on well-defined affective models of different complexity - polarity to distinguish positive and negative sentiment, for example, or more nuanced models to capture expressions of human emotions. When used to measure communication success, even the most granular affective model combined with sophisticated machine learning approaches may not fully capture an organisation's strategic positioning goals. Such goals often deviate from the assumptions of standardised affective models. While certain emotions such as Joy and Trust typically represent desirable brand associations, specific communication goals formulated by marketing professionals often go beyond such standard dimensions. For instance, the brand manager of a television show may consider fear or sadness to be desired emotions for its audience. This article introduces expansion techniques for affective models, combining common and commonsense knowledge available in knowledge graphs with language models and affective reasoning, improving coverage and consistency as well as supporting domain-specific interpretations of emotions. An extensive evaluation compares the performance of different expansion techniques: (i) a quantitative evaluation based on the revisited Hourglass of Emotions model to assess performance on complex models that cover multiple affective categories, using manually compiled gold standard data, and (ii) a qualitative evaluation of a domain-specific affective model for television programme brands. The results of these evaluations demonstrate that the introduced techniques support a variety of embeddings and pre-trained models. The paper concludes with a discussion on applying this approach to other scenarios where affective model resources are scarce.