Adults with autism face many difficulties when finding employment, such as struggling with interviews and needing accommodating environments for sensory issues. However, autistic adults also have unique skills to contribute to the workplace that companies have recently started to seek after, such as close attention to detail and trustworthiness. To work around these difficulties and help companies find the talent they are looking for we have developed a job-matching system. Our system is based around the stable matching of the Gale-Shapley algorithm to match autistic adults with employers after estimating how both adults with autism and employers would rank the other group. The system also uses filtering to approximate a stable matching even with a changing pool of users and employers, meaning the results are resistant to change as the result of competition. Such a system would be of benefit to both autistic adults and employers and would advance knowledge in recommendation systems that match two parties.

Indiana University-Purdue University Indianapolis (IUPUI) / This research proposed a dynamic recommendation system for a social learning environment entitled CourseNetworking (CN). The CN provides an opportunity for the users to satisfy their academic requirement in which they receive the most relevant and updated content. In our research, we extracted some implicit and explicit features from the system, which are the most relevant user feature and posts features. The selected features are used to make a rating scale between users and posts so that represent the link between user and post in this learning management system (LMS). We developed an algorithm which measures the link between each user and post for the individual. To achieve our goal in our system design, we applied natural language processing technique (NLP) for text analysis and applied various classi cation technique with the aim of feature selection. We believe that considering the content of the posts in learning environments as an impactful feature will greatly affect to the performance of our system. Our experimental results demonstrated that our recommender system predicts the most informative and relevant posts to the users. Our system design addressed the sparsity and cold-start problems, which are the two main challenging issues in recommender systems.

This study examines the design and trustworthiness factors of recommendation systems forexperience goods in the e-commerce industry. Experience goods are products that involvesensory experiences and pose challenges for consumers to assess and select online. Theresearch adopts a mixed method approach, combining exploratory and interpretive researchmethods to gain insights into users' interpretations and meanings attached to their experiences.The methodology includes analyzing publications, conducting a survey, and objectivelydocumenting recommendation systems in the alpine industry. The survey collects opinions fromparticipants who have used various recommendation systems, covering aspects such as usermodel, item model, recommendation algorithm, user interface, evaluation, and trustworthiness.A thematic analysis is employed to identify patterns and meaningful themes in the data. Thefindings emphasize the importance of understanding user preferences, balancingrecommendations, improving accuracy, enhancing interface usability, incorporating feedback,and addressing recommendation diversity to enhance trustworthiness. A hybrid filteringapproach with feature-based systems and integrated behavior-based techniques is identified aseffective. While the survey's convenience sampling and limited sample size may limitgeneralizability, the findings provide insights for designing effective recommendation systems forexperience goods in e-commerce. By considering the strengths and limitations of differenttechniques, vendors can create systems that assist customers in purchasing these uniqueproducts. However, recommendation systems should be viewed as a valuable tool rather thanthe sole determinant in purchase decisions for alpine equipment. Further research with a largerand more diverse sample is recommended to validate the findings and improve generalizability.

Sentiment analysis is a burgeoning field in natural language processing used to extract and categorize opinion in evaluative documents. We look at recommendation letters, which pose unique challenges to standard sentiment analysis systems. Our dataset is eighteen letters from applications to UMass Worcester Memorial Medical Center’s residency program in Obstetrics and Gynecology. Given a small dataset, we develop a method intended for use by domain experts to systematically explore their intuitions about the topical make-up of documents on which they make critical decisions. By leveraging WordNet and the WordNet Propagation algorithm, the method allows a user to develop topic seed sets from real data and propagate them into robust lexicons for use on new data. We show how one pass through the method yields useful feedback to our beliefs about the make-up of recommendation letters. At the end, future directions are outlined which assume a fuller dataset.

The use of smartphone applications has taken over the way people interact with the world. The design of an application has become an important aspect in keeping the user engaged [22]. People are looking for applications that are easy to use and will get the job done. This thesis focuses on the design of a mobile application for iOS that recommends wine in various retail locations that match a user’s taste preferences. The goals of this thesis are to design an iOS application that recommends wine to consumers, improve upon the wine recommendation algorithms by acquiring more customer data, and analyze the market for consumer and retail need for such a wine recommendation system. The mobile implementation developed for this thesis will be used by a startup based in San Luis Obispo called The Bottlefly. The application will supplement a similar in-store kiosk version to reach wine consumers outside of retail locations in hopes of bringing them into retail locations to purchase wine. Multiple studies are presented to show the results of acquiring customer data for the wine recommendation system as well as user interface usability studies to acquire data about the usability of the application. Usability factors such as ease of use, application completeness, and willingness to use are measured and analyzed in this thesis. The results will help propel the application forward to make sure it meets customer expectations in order to get it ready for production in retail locations and the App Store.

Top-N recommender systems automatically recommend N items for users from huge amounts of products. Personalized Top-N recommender systems have great impact on many real world applications such as E-commerce platforms and social networks. Sometimes there is no rating information in user-item feedback matrix but only implicit purchase or browsing history, that means the user-item feedback matrix is a binary matrix, we call such feedbacks as implicit feedbacks. In our work we try to learn Top-N recommender systems with implicit feedbacks. First, we design a heterogeneous loss function to learn the model. Second, we incorporate item side information into recommender systems. We formulate a low-rank constraint minimization problem and give a closed-form solution for it. Third, we also use item side information to learn recommender systems. We use gradient descent method to learn our model. Most existing methods produce personalized top-N recommendations by minimizing a specific uniform loss such as pairwise ranking loss or pointwise recovery loss. In our first model, we propose a novel personalized Top-N recommendation approach that minimizes a combined heterogeneous loss based on linear self-recovery models. The heterogeneous loss integrates the strengths of both pairwise ranking loss and pointwise recovery loss to provide more informative recommendation predictions. We formulate the learning problem with heterogeneous loss as a constrained convex minimization problem and develop a projected stochastic gradient descent optimization algorithm to solve it. Most previous systems are only based on the user-item feedback matrix. In many applications, in addition to the user-item rating/purchase matrix, item-based side information such as product reviews, book reviews, item comments, and movie plots can be easily collected from the Internet. This abundant item-based information can be used for recommendation systems. In the second model, we propose a novel predictive collaborative filtering approach that exploits both the partially observed user-item recommendation matrix and the item-based side information to produce top-N recommender systems. The proposed approach automatically identifies the most interesting items for each user from his or her non-recommended item pool by aggregating over his or her recommended items via a low-rank coefficient matrix. Moreover, it also simultaneously builds linear regression models from the item-based side information such as item reviews to predict the item recommendation scores for the users. The proposed approach is formulated as a rank constrained joint minimization problem with integrated least squares losses, for which an efficient analytical solution can be derived. In the third model, we also propose a joint discriminative prediction model that exploits both the partially observed user-item recommendation matrix and the item-based side information to build top-N recommender systems. This joint model aggregates observed user-item recommendation activities to predict the missing/new user-item recommendation scores while simultaneously training a linear regression model to predict the user-item recommendation scores from auxiliary item features. We evaluate the proposed approach on a variety of recommendation tasks. The experimental results show that the proposed joint model is very effective for producing top-N recommendation systems. / Computer and Information Science

Background: Tuberculosis (TB) is the number one infectious disease killer in the world. TB is both preventable, and curable. Since 1997, the World Health Organization’s (WHO) Global TB (GTB) Programme has released evidence-informed publications to guide member states. In their EndTB strategy, the WHO set a mandate to eradicate TB by 2035, in part by intensifying TB research and innovation. As an effort towards this goal, this project applies evidence mapping methodologies to published WHO TB recommendations, in an innovative process called “recommendation mapping” (RM). Objectives: The prime objective of RM is to allow guideline developers and key stakeholders to identify gaps and clusters of recommendations across publications, serve as an instrumental tool in the sequence of guideline development (from intelligent priority setting, to the assembly of final recommendations) and increase the accessibility of key guideline components. The secondary objective of this work is to poise guideline components for live update and refinement in a rapidly learning health system. Methods: In this mixed methods study, a methodological framework for mapping guideline components is proposed, with both a quantitative and narrative assessment of raw data and final map outputs. A qualitative analysis from the perspective of key stakeholders, policy-makers, researchers and WHO-GTB liaisons working in guideline development is also included. For the methodological piece, all publications containing WHO TB recommendations were eligible for the mapping exercise. Each recommendation was extracted according to all subdomains of their PICO backbone. Subsections of recommendations are coded using existing ontologies (SNOMED-CT, ATC, ICD-11). A centralized database containing extracted and coded recommendations was then presented in an online and interactive schematic. For the qualitative assessment of palatability of this approach within the organization, semi-structured interviews and a survey was delivered to eligible participants at two Guideline Development Group meetings for WHO tuberculosis treatment and screening guidelines. Results: The notable result of this work is the development, refinement and application of recommendation mapping methodologies. 20 WHO-GTB guidelines underwent an application of the novel recommendation methodologies proposed in this thesis to create an interactive map, and a searchable database. In-depth interviews and survey results with 21 participants (WHO GTB staff, WHO TB- guideline development group members and technical experts) pointed to concerns in the current accessibility and organization of WHO-GTB guidelines. Conclusions: Recommendation mapping may have utility in charting the terrain of recommendations, inform priority setting, and provide a scaffold for the future transition to living guidelines. / Thesis / Master of Public Health (MPH) / The World Health Organization (WHO) issues guidelines to help clinicians, policy-makers, and researchers make informed decisions in their work. Guidelines contain recommendations that can be thought of as bottom-line answers to the questions we ask the scientific literature (based on the evidence available to us today). The WHO’s Tuberculosis (TB) Department is partaking in a novel digital reorganization of their guideline recommendations using the evidence-mapping methods proposed in this thesis. This thesis uses the principles of evidence mapping to create recommendation maps that, like any map, chart the landscape in a given domain (in this case, TB recommendations). The recommendation map will help guide the WHO in setting priorities for future research and guideline development.

A frequent goal of a researcher is to publish his/her work in appropriate conferences and journals. With a large number of options for venues in the microdomains of every research discipline, the issue of selecting suitable locations for publishing cannot be underestimated. Further, the venues diversify themselves in the form of workshops, symposiums, and challenges. Several publishers such as IEEE and Springer have recognized the need to address this issue and have developed journal recommenders. In this thesis, our goal is to design and develop a similar recommendation system for the ACM dataset. We view this recommendation problem from a classification perspective. With the success of deep learning classifiers in recent times and their pervasiveness in several domains, we modeled several 1D Convolutional neural network classifiers for the different venues. When given some submission information like title, keywords, abstract, etc. about a paper, the recommender uses these developed classifier predictions to recommend suitable venues to the user. The dataset used for the project is the ACM Digital Library metadata that includes textual information for research papers and journals submitted at various conferences and journals over the past 60 years. We developed the recommender based on two approaches: 1) A binary CNN classifier per venue (single classifiers), and 2) Group CNN classifiers for venue groups (group classifiers). Our system has achieved a MAP of 0.55 and 0.51 for single and group classifiers. We also show that our system has a high recall rate. / Master of Science / A frequent goal of a researcher is to publish his/her research work in the form of papers and journals at recognized publication conferences and journals. Conferences limit the number of pages in a submission, whereas journals tend to be flexible with the length. In general, academic conferences are held annually, while journals have a submission cut off date on a monthly/trimonthly or so basis. These conferences and journals are publication venues. With a large number of options for venues in the microdomains of every research discipline, the issue of selecting suitable locations for publishing is a complicated task. Further, the venues diversify themselves in the form of workshops, symposiums, and challenges. Submitting a work to the wrong venue often leads to a rejection. Every author who is about to publish faces this question of ``Where can I publish my work so that it gets accepted?". This thesis is an attempt to address this question through a recommendation system. Recommendation systems help us in the decision making process. A well-known example is the ``Customers who bought this also bought item y'' message we find in eCommerce websites. These systems help users navigate a product catalog better to address their needs. The goal of this thesis is to develop one such recommendation system that can help researchers to choose venues. When an author is about to publish, they structure their paper/journal in the form of a research title, brief abstract, relevant keywords in the paper, and a detailed explanation of the research carried out. Our system can take any of these as input and suggest appropriate venues based on the submission content. The dataset used for the project is the ACM Digital Library metadata. We developed the recommender using deep learning techniques. Our system can be helpful for finding a single best venue, or a group of suitable venues.

Online group-buying is one of the popular online business models recently. Both the initiator and participants hope to recruit more consumers to join order to aggregate larger orders and thus get cheaper price. Traditionally, consumers always invite their friends or families to join group-buying in order to collect more orders. Hope the relationship could affect their behavior. As the communication and coordination through the Internet are getting more convenient, it is easy and popular to recruit friends in larger range to join group-buying via e-mail. Further, the increasing virtual communities result from that, members have same interest, concern, and needs. It is quite possible that the members have same needs and therefore initiate a group-buying activity to fulfill many members¡¦ needs. Since information sharing is a major activity between members of virtual communities, the degree of the interactions will impact the tie strength between them. If members can send peer-to-peer recommendation email to other members who may be interested in the group-buying transaction, it may improve the group-buying performance. In addition, marketing via e-mail is getting common. The different marketing appeal results in different effect. Rational appeals focus on product itself while emotional appeal makes consumer¡¦s feeling change. The purpose of this research is to explore the difference in advertisement attitude between consumers clicking the peer-to-peer recommendation e-mail and consumers not clicking it. We also examined if these two groups have different cognition of tie strength with the e-mail sender. The result shows the group clicking the recommendation mail has better advertisement attitude than the group not clicking. Further, emotional appeal induces the subjects¡¦ better cognition of reliability of the appeal

Recommendation techniques are developed to discover user¡¦s real information need among large amounts of information. Recommendation systems help users filter out information and attempt to present those similar items according to user¡¦s tastes. In our work, we focus on discussion threads recommendation in the tourism domain. We assume that when users have traveling information need, they will try to search related information on the websites. In addition to browsing others suggestions and opinions, users are allowed to express their need as a question. Hence, we focus on recommending users previous discussion threads that may provide good answers to the users¡¦ questions by considering the question input as well as their browsing records. We propose a model, which consists of four perspectives: goal similarity, content similarity, freshness and quality. To validate and the effectiveness of our model on recommendation performance, we collected 14348 threads from TripAdvisor.com, the largest travel website, and recruited ten volunteers, who have interests in the tourism, to verify our approach. The four perspectives are utilized by two methods. The first is Question-based method, which makes use of content similarity, freshness and quality and the second is Session-based method, which involves goal similarity. We also integrate the two methods into a hybrid method. The experiment results show that the hybrid method generally has better performance than the other two methods.