Hybrid Variational Autoencoder for Recommender Systems

E-commerce platforms heavily rely on automatic personalized recommender systems, e.g., collaborative filtering models, to improve customer experience. Some hybrid models have been proposed recently to address the deficiency of existing models. However, their performances drop significantly when the dataset is sparse. Most of the recent works failed to fully address this shortcoming. At most, some of them only tried to alleviate the problem by considering either user side or item side content information. In this article, we propose a novel recommender model called Hybrid Variational Autoencoder (HVAE) to improve the performance on sparse datasets. Different from the existing approaches, we encode both user and item information into a latent space for semantic relevance measurement. In parallel, we utilize collaborative filtering to find the implicit factors of users and items, and combine their outputs to deliver a hybrid solution. In addition, we compare the performance of Gaussian distribution and multinomial distribution in learning the representations of the textual data. Our experiment results show that HVAE is able to significantly outperform state-of-the-art models with robust performance.

Social-Spatial Group Queries with Keywords

Social networks with location enabling technologies, also known as geo-social networks, allow users to share their location-specific activities and preferences through check-ins. A user in such a geo-social network can be attributed to an associated location (spatial), her preferences as keywords (textual), and the connectivity (social) with her friends. The fusion of social, spatial, and textual data of a large number of users in these networks provide an interesting insight for finding meaningful geo-social groups of users supporting many real-life applications, including activity planning and recommendation systems. In this article, we introduce a novel query, namely, Top- k Flexible Socio-Spatial Keyword-aware Group Query (SSKGQ), which finds the best k groups of varying sizes around different points of interest (POIs), where the groups are ranked based on the social and textual cohesiveness among members and spatial closeness with the corresponding POI and the number of members in the group. We develop an efficient approach to solve the SSKGQ problem based on our theoretical upper bounds on distance, social connectivity, and textual similarity. We prove that the SSKGQ problem is NP-Hard and provide an approximate solution based on our derived relaxed bounds, which run much faster than the exact approach by sacrificing the group quality slightly. Our extensive experiments on real data sets show the effectiveness of our approaches in different real-life settings.

Identification of Reproductive Health Monitoring Indicators in Iran

Introduction As a result of recent demographic changes, Iran has revised its reproductive health programs. To respond to the essential need for monitoring the new programs and policies, this study aimed to identify tailored, appropriate, and measurable RH indicators in the Iranian context, using available evidence and international indicators.Method This is an applied mixed-methods research, which was conducted in four phases: Identification of goals of RH policies and programs, scoping review of the RH indicators in the literature, developing and ranking the identified indicators, and finalization of indicators. Qualitative content analysis was used to analyze the textual data of the documents and policies. We analyzed the studies in the scoping review by narrative synthesis. The final indicators were selected through the consensus of experts, with a cut-off point of 75%. Result We identified 689 indicators through document analysis and scoping review. After three round of screening, a total of 37 RH indicators were finalized. The first five indicators with the highest score were: total fertility rate, population under 15 years, total population, population aged 65 years and older, and age-specific fertility rate.Conclusion: The nature and number of indicators for monitoring and evaluation of reproductive health programs might vary at different organizational levels; hence the need to develop specific indicators for each level is pivotal. In addition, the need for collection, processing and dissemination of reliable data for evaluation of these programs is essential.

Quantitative assessment of information quality in textual sources for landslide inventories

Landslide research chiefly relies on digital inventories for a multitude of spatial, temporal, and/or process analyses. In respect thereof, many landslide inventories are populated with information from textual documents (e.g., news articles, technical reports) due to effectiveness. However, information detail can vary greatly in these documents and the question arises whether such textual information is suitable for landslide inventories. The present work proposes to define the usefulness of textual source types as a probability to find landslide information, weighted with adaptable parameter requirements. To illustrate the method with practical results, a German landslide dataset has been examined. It was found that three combined source types (administrative documents, expert opinions, and news articles) give an 89 % chance to detect useful information on three defined parameters (location, date, and process type). In conclusion, the definition of usefulness as a probability makes it an intuitive, quantitative measure that is suitable for a wide range of applicants. Furthermore, a priori knowledge of usefulness allows for focusing on a few source types with the most promising outcome and thus increases the effectiveness of textual data acquisition and digitalisation for landslide inventories.

Fuzzy Inference Propelled Sentence Ranking for Extractive Summary Generation

Natural language serves as an impeccable tool for the appropriate representation of knowledge among individuals. Owing to the varying representation of the same knowledge base and the perpetual growth of the World Wide Web, the need to uncover an effective method to condense available textual data without significantly dampening the implied information is paramount. In an attempt to solve the need for effectively condensing textual data, the paper proposes a system which is capable of mimicking the human brain's approach to process Natural Language Fuzzy Logic. The system is subjected to both intrinsic and extrinsic evaluation and the results are compared against two other text summarizers - Auto summarize Tool and SweSum using the CNN Corpus Dataset. The Relevance Prediction Measure, F1 Score and Recall results suggest the applicability of Fuzzy Reasoning in text summarization and through evaluation, it can be inferred that proposed system has successfully tried to mimic the process of summary generation by the human brain.