Search for: All records

An overall rating cannot reveal the details of user’s preferences toward each feature of a product. One widespread practice of e-commerce websites is to provide ratings on predefined aspects of the product and user-generated reviews. Most recent multi-criteria works employ aspect preferences of users or user reviews to understand the opinions and behavior of users. However, these works fail to learn how users correlate these information sources when users express their opinion about an item. In this work, we present Multi-task & Multi-Criteria Review-based Rating (MMCRR), a framework to predict the overall ratings of items by learning how users represent their preferences when using multi-criteria ratings and text reviews. We conduct extensive experiments with three real-life datasets and six baseline models. The results show that MMCRR can reduce prediction errors while learning features better from the data. 

Classifying multivariate time series (MTS), which record the values of multiple variables over a continuous period of time, has gained a lot of attention. However, existing techniques suffer from two major issues. First, the long-range dependencies of the time-series sequences are not well captured. Second, the interactions of multiple variables are generally not represented in features. To address these aforementioned issues, we propose a novel Cross Attention Stabilized Fully Convolutional Neural Network (CA-SFCN) to classify MTS data. First, we introduce a temporal attention mechanism to extract long- and short-term memories across all time steps. Second, variable attention is designed to select relevant variables at each time step. CA-SFCN is compared with 16 approaches using 14 different MTS datasets. The extensive experimental results show that the CA-SFCN outperforms state-of-the-art classification methods, and the cross attention mechanism achieves better performance than other attention mechanisms.

 

Most e-commerce websites (e.g., Amazon and TripAdvisor) show their users an initial set of useful product reviews. These reviews allow users to form a general idea about the product’s characteristics. The usefulness of a review is mainly based on a score that the website users provide. Studies have shown that this score is not a good indicator of a review’s actual helpfulness. Nonetheless, most past works still use it to classify a review as helpful or not. With the growing number of reviews, finding those helpful ones is a challenging task. In this work, we propose NovRev, a new unsupervised approach to recommend a personalized subset of unread useful reviews for those users looking to increase their knowledge about a product. NovRev considers an initial set of reviews as a context and recommends reviews that increase the product’s information. We have extensively tested NovRev against five baseline methods, using eight real-life datasets from different product domains. The results show that NovRev can recommend novel, relevant, and diverse reviews while covering more information about the product. 

ABSTRACT Elizabethkingia anophelis is an emerging global multidrug-resistant opportunistic pathogen. We assessed the diversity among 13 complete genomes and 23 draft genomes of E. anophelis strains derived from various environmental settings and human infections from different geographic regions around the world from 1950s to the present. Putative integrative and conjugative elements (ICEs) were identified in 31/36 (86.1%) strains in the study. A total of 52 putative ICEs (including eight degenerated elements lacking integrases) were identified and categorized into three types based on the architecture of the conjugation module and the phylogeny of the relaxase, coupling protein, TraG, and TraJ protein sequences. The type II and III ICEs were found to integrate adjacent to tRNA genes, while type I ICEs integrate into intergenic regions or into a gene. The ICEs carry various cargo genes, including transcription regulator genes and genes conferring antibiotic resistance. The adaptive immune CRISPR-Cas system was found in nine strains, including five strains in which CRISPR-Cas machinery and ICEs coexist at different locations on the same chromosome. One ICE-derived spacer was present in the CRISPR locus in one strain. ICE distribution in the strains showed no geographic or temporal patterns. The ICEs in E. anophelis differ in architecture and sequence from CTnDOT, a well-studied ICE prevalent in Bacteroides spp. The categorization of ICEs will facilitate further investigations of the impact of ICE on virulence, genome epidemiology, and adaptive genomics of E. anophelis . IMPORTANCE Elizabethkingia anophelis is an opportunistic human pathogen, and the genetic diversity between strains from around the world becomes apparent as more genomes are sequenced. Genome comparison identified three types of putative ICEs in 31 of 36 strains. The diversity of ICEs suggests that they had different origins. One of the ICEs was discovered previously from a large E. anophelis outbreak in Wisconsin in the United States; this ICE has integrated into the mutY gene of the outbreak strain, creating a mutator phenotype. Similar to ICEs found in many bacterial species, ICEs in E. anophelis carry various cargo genes that enable recipients to resist antibiotics and adapt to various ecological niches. The adaptive immune CRISPR-Cas system is present in nine of 36 strains. An ICE-derived spacer was found in the CRISPR locus in a strain that has no ICE, suggesting a past encounter and effective defense against ICE. 

Skyline queries are used to find the Pareto optimal solution from datasets containing multi-dimensional data points. In this paper, we propose a new type of skyline queries whose evaluation is constrained by a multi-cost transportation network (MCTN) and whose answers are off the network. This type of skyline queries is useful in many applications. For example, a person wants to find an apartment by considering not only the price and the surrounding area of the apartment, but also the transportation cost, time, and distance between the apartment and his/her work place. Most existing works that evaluate skyline queries on multi-cost networks (MCNs), which are either MCTNs or road networks, find interesting objects that locate on edges of the networks. Formally, our new type of skyline queries takes as input an MCTN, a query point q, and a set of objects of interest D with spatial information, where q and the objects in D are off the network. The answers to such queries are objects in D that are not dominated by other D objects when considering the multiple attributes of these objects and the multiple network cost from q to the solution objects. To evaluate such queries, we propose an exact search algorithm and its improved version by implementing several properties. The space of the exact skyline solutions is huge and can easily reach the order of thousands and incur long evaluation time. We further design much more efficient heuristic methods to find approximate solutions. We run extensive experiments using both real and synthetic datasets to test the effectiveness and efficiency of our proposed approaches. The results show that the exact search algorithm can be dramatically improved by utilizing several properties. The heuristic approaches to find approximate answers can largely reduce the query time and retrieve results that are comparable to the exact solutions.