Item Ratings Research Articles

Examining Higher Education Instructor Perceptions of Roles and Competencies in Online Teaching

Online instructors adopt various roles and perform various competencies in the design and delivery of online courses. In this study, online instructor roles are categorized into eight types including Subject Matter Expert, Course Designer and Developer, Course Facilitator, Course Manager, Advisor/Mentor, Assessor/Evaluator, Technology Expert, and Lifelong Learner. Through survey-based research with 141 online instructors, this study examines competencies that online instructors perform based on various roles. When rating competencies, overall categorical means for all the roles were rated above 4.00, which showed that they used all these roles. The highest rated items and lowest rated items are discussed in addition to the connection between research and practice in online teaching. Online instructors who participate in training and who collaborate with instructional designers rated the frequency with which they perform the competencies to be higher. This study has implications for online instructors, instructional designers, and administrators who design and deliver online learning and offer professional development for online instructors.

LBCF: A Link-Based Collaborative Filtering for Overfitting Problem in Recommender System

Recommender system (RS) suggests relevant objects to generate personalized service and minimize information overload issue. User-based collaborative filtering (UBCF) plays a dominant role in practical RSs. However, traditional UBCF suffers from a recommendation overfitting problem, i.e., recommendations generated by UBCF usually concentrate on popular items, resulting in lower diversity. In addition, UBCF cannot maintain a reasonable tradeoff between the accuracy and diversity of recommendations because raising the diversity is often accompanied by a decrease in accuracy. In this article, we propose a novel approach, namely link-based collaborative filtering, to enhance the recommendation accuracy and diversity simultaneously without employing additional complex information. First, a user–item bipartite network is constructed based on the user–item rating matrix of RSs. Then, a global–local weighted bipartite modularity is presented to conduct link partition so that links with the same community can not only be relatively denser but also own the same characteristic. Furthermore, redundant links are removed from each community by utilizing a link reduction algorithm so that neighborhood of a target user can be selected according to the more efficient nonredundant links. Finally, rating prediction is executed based on the rating information of neighborhood. Also, items owning the highest predicted rating scores will be recommended to the target user. Experimental results from three real datasets of RSs suggest that, without taking advantage of special additional data, our proposed approach outperforms the state-of-the-art studies and is able to generate personalized recommendations with satisfying accuracy and diversity simultaneously.

Cognitive Aid for Anesthetic Preparation in An Emergency Situation: A Simulation-Based Study.

When anesthesia checklists and preparations are performed urgently, omissions may occur and be deleterious to the patient. The aim of this study was to evaluate in simulation the interest of a cognitive aid to effectively prepare an anesthetic room for an emergency. In a prospective single-center simulation-based study, 32 anesthesia residents had to prepare an anesthetic room in an emergency scenario, without cognitive aid in the first phase. Three months later (phase 2), they were randomly assigned to receive a cognitive aid (aid group) or no additional aid (control) and were involved in the same scenario. The primary outcome was the validation rate of each essential item in the first 5 min in phase 2. Eight items were significantly more frequently completed in the first 5 min in the aid group in phase 2 (vs. phase 1), compared with two only in the control group. However, there were no significant differences in the overall number of completed items between the two groups, as both groups completed significantly more items in phase 2, either in the first 5 min (19 (14–23) vs. 13 (9–15) in phase 1 for all residents, p < 0.001) or without time limit. Preparation times were reduced in phase 2 in both groups. In conclusion, the use of a cognitive aid allowed anesthesia residents to complete some safety items of a simulated urgent anesthesia preparation more frequently. In addition, despite daily clinical experience, a single simulation session improved anesthesia preparation and reduced the preparation time with or without cognitive aid.

Studying Antillean manatees’ (Trichechus manatus manatus) temperament in zoological parks: exploration of boldness, sociality and reactivity to humans

Temperament, i.e. consistent inter-individual differences in behavior, has been documented in many species. Especially, boldness-shyness continuum is the ecologically relevant trait most frequently used to describe temperament among species. Reactions along the bold-shy axis demonstrate an organism’s ability to cope with environmental conditions and reflect its response to environmental stress. Our study investigated the existence of boldness-shyness continuum in 16 captive Antillean manatees (Trichechus manatus manatus) housed in 2 zoological parks. We first aimed to establish a valid methodology to measure boldness and shyness using behavioral tests. Additionally, our study explored 2 other potential temperament dimensions: sociality and reactivity to humans, and their association with boldness. Finally, since animals were kept under human care, we investigated human perception on manatees’ temperament. We conducted behavioral tests i.e. novelty tests involving novel objects, a novel sound, and an unfamiliar human. We recorded behaviors related to exploration as well as social interactions (i.e. physical proximity, social contacts) and we assessed manatees’ reactivity to humans through a participation score during hand-feeding sessions. Finally, we collected keepers’ subjective impression about individuals’ temperament through traits rating method. To our knowledge, this is the first paper that tested manatees’ reactions to different novel stimuli in order to determine their temperament. Our results first allowed us to identify the most relevant stimuli and variables to measure boldness and to classify individuals as bolder or shier. Individuals characterized as bolder displayed significantly more approach, longer approach and more contact toward stimuli than shier individuals. Moreover, boldness score was significantly associated with physical proximity with conspecifics, as well as with the participation score to hand-feeding sessions in one zoo. Finally, surveys completed by keepers were reliable and positive associations were found between rated items and behavioral observations. These findings support that as many species, manatees demonstrate boldness-shyness traits. More investigations on a larger sample size are needed to explore manatees’ temperament dimensions, and how these traits could modulate their interactions with their conspecifics and their environment. Studying manatees’ temperament could help improving the management and welfare of the species under human care. It could also facilitate their management in wildlife reserves by understanding manatees’ movements and evaluate the anthropogenic impact on wild populations.

Web crawling based context aware recommender system using optimized deep recurrent neural network

Recommendation systems are obtaining more attention in various application fields especially e-commerce, social networks and tourism etc. The top items are recommended based on the ability of recommender system which predict the future preference out of the available items. Because of the internet, the people in the current society has too many options that’s why the recommendation system is very essential. The recommendation is achieved by the particular users who predict the ratings for numerous items and recommend those items to other users. Majorly, content and collaborative filtering techniques are employed in typical recommendation systems to find user preferences and provide final recommendations. But, these systems commonly lacks to take growing user preferences in various contextual factors. Context aware recommendation systems consider various contextual parameters into account and attempt to catch user preferences appropriately. The majority of the work in the recommender system domain focuses on increasing the recommendation accuracy by employing several proposed approaches where the main motive remains to maximize the accuracy of recommendations while ignoring other design objectives, such as a user’s an item’s context. Therefore, in this paper an effective deep learning based context aware recommendation model is proposed which can be act as an efficient recommender system by showing minimum error during recommendation. Initially, the dataset is pre-processed using Natural Language Tool Kit (NLTK) in Python platform. After pre-processing, the TF–IDF and word embedding model is used for every pre-processed reviews to extract the features and contextual information. The extracted feature is considered as an input of density based clustering to group the negative, neutral and positive sentiments of user reviews. Finally, deep recurrent neural Network (DRNN) is employed to get the most preferable user from every cluster. The recurrent neural network model parameter values are initialized through the fitness computation of Bald Eagle Search (BES) algorithm. The proposed model is implemented using NYC Restaurant Rich Dataset using Python programming platform and performance is evaluated based on the metrics of accuracy, precision, recall and compared with existing models. The proposed recommendation model achieves 99.6% accuracy which is comparatively higher than other machine learning models.

The Relationship Between Reserve Items and Exchange Rate in Malaysia: Evidence from Non-linear Model

The objectives of this study are first, to examine a nonlinear long-run relationship between exchange rate and reserve and related item in Malaysia. Second, to examine the causal relationship between exchange rate, reserve and related item, money supply, interest rate, foreign direct investment, and price in Malaysia. Third, to forecast the relationship between exchange rate, reserve and related item, money supply, interest rate, foreign direct investment, and price in Malaysia for the next 50 years. To achieve our objective, we employed the nonlinear autoregressive distributed lag model (NARDL), Toda Yamamoto causality test and variance decomposition. We utilize a time series dataset from 1982 to 2018. We find reserve items are significant to have a negative relationship with exchange rate. However, it limits to only a decrease in reserve items will encourage the Ringgit to be depreciated. On the other hand, reserve items are expected to influence more in exchange rate fluctuation in the next 50 years. On the other hand, reserve items are significant to cause FDI inflow in Malaysia.

Novel Multidimensional Collaborative Filtering Algorithm Based on Improved Item Rating Prediction

Current data has the characteristics of complexity and low information density, which can be called the information sparse data. However, a large amount of data makes it difficult to analyse sparse data with traditional collaborative filtering recommendation algorithms, which may lead to low accuracy. Meanwhile, the complexity of data means that the recommended environment is affected by multiple dimensional factors. In order to solve these problems efficiently, our paper proposes a multidimensional collaborative filtering algorithm based on improved item rating prediction. The algorithm considers a variety of factors that affect user ratings; then, it uses the penalty to account for users’ popularity to calculate the degree of similarity between users and cross-iterative bi-clustering for the user scoring matrix to take into account changes in user’s preferences and improves on the traditional item rating prediction algorithm, which considers user ratings according to multidimensional factors. In this algorithm, the introduction of systematic error factors based on statistical learning improves the accuracy of rating prediction, and the multidimensional method can solve data sparsity problems, enabling the strongest relevant dimension influencing factors with association rules to be found. The experiment results show that the proposed algorithm has the advantages of smaller recommendation error and higher recommendation accuracy.

An expert survey to assess the current status and future challenges of energy system analysis

Decision support systems like computer-aided energy system analysis (ESA) are considered one of the main pillars for developing sustainable and reliable energy strategies. Although today's diverse tools can already support decision-makers in a variety of research questions, further developments are still necessary. Intending to identify opportunities and challenges in the field, we classify modelling topics into modelling capabilities (32), methodologies (15), implementation issues (15) and management issues (7) from an extensive literature review. Based on a quantitative expert survey of energy system modellers (N = 61) mainly working with simulation and optimisation models, the Status of Development and the Complexity of Realisation of those modelling topics are assessed. While the rated items are considered to be more complex than actually represented, no significant outliers are determinable, showing that there is no consensus about particular aspects of ESA that are lacking development. Nevertheless, a classification of the items in terms of a specially defined modelling strategy matrix identifies capabilities like land-use planning patterns, equity and distributional effects and endogenous technological learning as “low hanging fruits” for enhancement, as well as a large number of complex topics that are already well implemented. The remaining “tough nuts” regarding modelling capabilities include non-energy sector and social behaviour interaction effects. In general, the optimisation and simulation models differ in their respective strengths, justifying the existence of both. While methods were generally rated as quite well developed, combinatorial optimisation approaches, as well as machine learning, are identified as important research methods to be developed further for ESA.

A feasibility study of virtual group therapy to improve quality of life of cancer caregivers

Background: Cancer caregiving can negatively impact the quality of life (QOL) of the caregiver. In-person interventions for improving coping skills have been shown to be effective in improving QOL for caregivers. Objectives: This pilot project explored the feasibility and acceptability of a virtual group therapy intervention to improve short-term cancer caregiver QOL. Methods: Caregivers of cancer patients were enrolled in a structured multidisciplinary intervention of eight virtual group therapy sessions provided over four weeks between September 9, 2013 and November 17, 2014. Group sessions were led by trained facilitators and included components of physical therapy, occupational therapy, psychosocial education, cognitive-behavioral intervention, supportive discussion, spiritual reflection, and mindfulness therapy. Feasibility was based on acceptable number of recruited participants per session; acceptability was defined using attendance and 80% QOL completion rates. QOL domains and symptom burden were assessed using validated single items. Results: The 20 cancer caregivers who enrolled were mostly older (80% were ≥ 65 years), female (76.5%), married to the patient (88.2%), Caucasian (100%), and highly educated (100%). 60% attended one to five sessions, 15% attended six to eight sessions, and 25% attended no sessions. Thirty percent completed pre- and post- intervention ratings of QOL items. Significance of Results: Findings suggested that a virtual group therapy intervention is feasible for the cancer caregivers in this study. Although not statistically significant, the caregivers reported higher QOL and less symptom burden in multiple domains after participating in the virtual group therapy intervention.

Gynecologic Interstitial Brachytherapy Curriculum Using a Low-Cost Phantom With Ultrasound Workshop and a Treatment Planning Workshop is Feasible and Effective

<h3>Purpose/Objective(s)</h3> There is no standard simulation training geared towards interstitial brachytherapy (IS BT) for gynecologic malignancies. Most residents perform few IS cases during residency. However, IS BT is an integral part of care for locally advanced gynecologic malignancies. We developed and implemented a workshop curriculum for IS BT with: 1) lecture on equipment, workflow, and guidelines, 2) hands-on ultrasound-guided IS BT workshop, and 3) treatment planning workshop. <h3>Materials/Methods</h3> The cost of each gynecologic phantom was approximately $70; phantoms were composed of an acrylic vase cube, rigid PETG tubes, modeling clay, agar, glass microspheres, a custom molded cervix/uterus structure, and a water-filled condom to mimic the bladder. A hands-on ultrasound-guided IS BT workshop was performed with residents in two small groups, where they imaged the gynecologic phantom using a transabdominal ultrasound probe. Participants alternated roles of sonographer or brachytherapist, implanting the phantom using custom stainless-steel needles machined in-house. A second workshop with hands-on treatment planning was performed in planning software using a L-Q spreadsheet with the following objectives: coverage goal of D90, meeting D2cc constraints of bladder and rectum, and minimizing V200 to qualitatively involve a small volume. All residents filled out a pre- and post-survey with subjective (4 demographics; 12 Likert-style) and objective knowledge-based (13) questions about GYN IS-BT. Paired subjective data were compared with the Wilcoxon signed ranks test. Paired objective data were compared with the McNemar's test of paired proportions. All statistical analysis was performed with a 2-sided <i>P</i>-value at the 0.05 significance level. <h3>Results</h3> Four residents had prior IS BT experience. After the workshop, residents reported significantly improved comfort with skills and knowledge regarding GYN IS BT equipment and procedure, evaluating GYN anatomy using ultrasound, and CT Simulation, contouring, and plan review (median pre-session Likert score 2 [1-3] vs post session 3 [3-4], <i>P</i> < 0.05). The overall rate of percent correct items assessing objective radiation oncology knowledge at baseline was 44%, Following completion of the curriculum, the overall percent correct items was 88% (<i>P</i> < 0.0001). All residents "Agree" (2/7) and "Strongly Agree" (5/7) the session was an effective learning experience. Qualitative feedback primarily asked for more time with treatment planning and educational content beforehand. <h3>Conclusion</h3> Residents participating in phantom training with an ultrasound curriculum and a treatment planning session is both feasible and effective. Similar strategies should be considered to enhance and standardize radiation oncology education initiatives for gynecologic IS BT training.

Occurrence of Phthalates in Bottled Drinks in the Chinese Market and Its Implications for Dietary Exposure.

Ubiquitous occurrences of phthalic acid esters (PAEs) or phthalates in a variety of consumer products have been demonstrated. Nevertheless, studies on their occurrence in various types of bottled drinks are limited. In this study, fifteen PAEs were analyzed in six categories of bottled drinks (n = 105) collected from the Chinese market, including mineral water, tea drinks, energy drinks, juice drinks, soft drinks, and beer. Among the 15 PAEs measured, DEHP was the most abundant phthalate with concentrations ranging from below the limit of quantification (LOQ) to 41,000 ng/L at a detection rate (DR) of 96%, followed by DIBP (DR: 88%) and DBP (DR: 84%) with respective concentration ranges of below LOQ to 16,000 and to 4900 ng/L. At least one PAE was detected in each drink sample, and the sum concentrations of 15 PAEs ranged from 770 to 48,004 ng/L (median: 6286 ng/L). Significant differences with respect to both PAE concentrations and composition profiles were observed between different types of bottled drinks. The median sum concentration of 15 PAEs in soft drinks was over five times higher than that detected in mineral water; different from other drink types. Besides DEHP, DBIP, and DBP, a high concentration of BMEP was also detected in a tea drink. The estimated daily dietary intake of phthalates (EDIdrink) through the consumption of bottled drinks was calculated based on the concentrations measured and the daily ingestion rates of bottled drink items. The EDIdrink values for DMP, DEP, DIBP, DBP, BMEP, DAP, BEEP, BBP, DCP, DHP, BMPP, BBEP, DEHP, DOP, and DNP through the consumption of bottled mineral water (based on mean concentrations) were 0.45, 0.33, 12.5, 3.67, 2.10, 0.06, 0.32, 0.16, 0.10, 0.09, 0.05, 0.81, 112, 0.13, and 0.20 ng/kg-bw/d, respectively, for Chinese adults. Overall, the EDIdrink values calculated for phthalates through the consumption of bottled drinks were below the oral reference doses suggested by the United States Environmental Protection Agency (U.S. EPA).

Usability and reproducibility of three tools to assess medical students and residents in emergency medicine.

It is critical to assess competency of medical students and residents in emergency medicine (EM) during undergraduate and graduate medical education. However, very few valid tools exist to assess both technical and nontechnical skills in the specific context of EM. Three Acute Care Assessment Tools (ACAT 1, 2, and 3) have been previously developed for three acute care conditions: cardiac arrest (1), coma (2), and acute respiratory failure (3). This study aimed to evaluate the reproducibility of the tools. The tool was tested using recorded videos of simulation sessions of fourth year medical students and first year residents in EM. Raters independently reviewed the videos two times in a 3-month interval, and interrater and intrarater reliability using intraclass correlation (ICC) was calculated. Secondary endpoints included the completeness rate and relevance of the ACAT. Sixty-two sessions were recorded and 48 videos analyzed (18 for CA and 15 for both respiratory failure and coma. The learners were residents in 32 (66%) of videos. Interrater reliability was excellent (ICC >0.9 for all three contexts) and so was the intrarater reliability (>0.88), both upon first review (month 0, M0) and at 3months (M3). The usability of the ACAT was good, with a completeness of the items that ranged from 96% to 100%. Only one item of the ACAT 1 had a relevance of 27%, as it could not be completed in 13 scenarios out of 18. The results demonstrate educators can evaluate students similarly utilizing video recordings of simulated medical scenario. The excellent completeness of the rated items advocated for good usability. The three ACATs can be utilized to assess for completeness of predefined tasks in three acute care broad scenario in a competency-based medical education framework.

User Curiosity Factor in Determining Serendipity of Recommender System

Recommender rystem (RS) is created to solve the problem by recommending some items among a huge selection of items that will be useful for the e-commerce users. RS prevents the users from being flooded by information that is irrelevant for them.Unlike information retrieval (IR) systems, the RS system's goal is to present information to the users that is accurate and preferably useful to them. Too much focus on accuracy in RS may lead to an overspecialization problem, which will decrease its effectiveness. Therefore, the trend in RS research is focusing beyond accuracy methods, such as serendipity. Serendipity can be described as an unexpected discovery that is useful. Since the concept of a recommendation system is still evolving today, formalizing the definition of serendipity in a recommendation system is very challenging.One known subjective factor of serendipity is curiosity. While some researchers already addressed curiosity factor, it is found that the relationships between various serendipity component as perceived by the users and their curiosity levels is still yet to be researched. In this paper, the method to determine user curiosity model by considering the variation of rated items was presented, then relation to serendipity components using existing user feedback data was validated. The finding showed that the curiosity model was related to some user-perceived values of serendipity, but not all. Moreover, it also had positive effect on broadening the user preference.

Expanding Availability of Speech-Generating Device Evaluation and Treatment to People With Amyotrophic Lateral Sclerosis (pALS) Through Telepractice: Perspectives of pALS and Communication Partners

Purpose To examine the experiences of people with ALS (pALS) and their communication partners (cALS) regarding receiving speech-generating device (SGD) evaluation and treatment via telepractice. Method Eight pALS along with a primary cALS participated in telepractice SGD evaluation and treatment with an augmentative and alternative communication (AAC) specialist and representatives from multiple SGD vendors. Participants were interviewed postevaluation and post-SGD training to examine their experiences. Mixed methods data were collected through Likert scale responses and qualitative interviews. Results Telepractice SGD evaluation and training were feasible and resulted in all pALS receiving SGDs they were able to use to communicate. In both Likert rating items and qualitative interviews, participants rated the telepractice experience very highly in terms of giving them access to AAC services via an AAC specialist that they would not have otherwise been able to access, and doing so in a format that was possible given their limitations in mobility, endurance, and caregiver availability. Suggestions for improving the telepractice experience were provided. Conclusions Telepractice should be considered as an option to provide vital SGD services to patients who are geographically remote, mobility impaired, unable to leave their home, experience fatigue with travel, or otherwise would not have access to these specialized services. Telepractice allows patients to preserve their time and energy for the assessment and treatment sessions, resulting in perhaps deeper and more frequent engagement in evaluation and training. Telepractice could serve as an alternative to outpatient, in-person evaluations, or be utilized in conjunction with in-person appointments. Supplemental Material https://doi.org/10.23641/asha.15094257.

Towards psychology-aware preference construction in recommender systems: Overview and research issues

User preferences are a crucial input needed by recommender systems to determine relevant items. In single-shot recommendation scenarios such as content-based filtering and collaborative filtering, user preferences are represented, for example, as keywords, categories, and item ratings. In conversational recommendation approaches such as constraint-based and critiquing-based recommendation, user preferences are often represented on the semantic level in terms of item attribute values and critiques. In this article, we provide an overview of preference representations used in different types of recommender systems. In this context, we take into account the fact that preferences aren’t stable but are rather constructed within the scope of a recommendation process. In which way preferences are determined and adapted is influenced by various factors such as personality traits, emotional states, and cognitive biases. We summarize preference construction related research and also discuss aspects of counteracting cognitive biases.

Вплив респіраторного компоненту на задоволеність фізичною терапією після кардіохірургічних втручань

The purpose of the study was to investigate the effect of additional respiratory physical therapy on the level of satisfaction with physical therapy among patients after cardiac surgery. Materials and methods. The study involved 150 patients. The patients were randomly divided into three groups: control (respiratory physical therapy was limited to cough), group of incentive spirometry (in classes with a physical therapist patients performed additional 3 approaches of 10 breaths through the simulator; they received recommendations for hourly performance of a similar number of cycles of exercise), a group of patients who received additional respiratory physical therapy with positive expiratory pressure in the form of exhalation into a bottle of water through a tube (number of repetitions and recommendations are similar to those received by the previous group). All groups underwent the same protocol of mobilization and use of therapeutic physical exercises in the procedure of therapeutic gymnastics. Performing exercises with breathing simulators began on the first postoperative day. Results and discussion. A questionnaire (17 questions) was used to assess the satisfaction with physical therapy. The survey was conducted on the seventh postoperative day. The answers were distributed on a 5-point Likert scale from 1 point to 5 points. The key preoperative indicators and time indicators of surgical interventions of the examined groups did not differ significantly. Analysis of satisfaction with physical therapy revealed a high level in all items of the questionnaire. Statistical differences between the groups were found in only three items of the questionnaire, but they were not related to either the use of equipment or diligence in treatment or the content of physical therapy. The overall score of the questionnaire did not differ in groups according to the Kraskel-Wallis criterion, and Me indicators (25%; 75%) were in the control group – 81 (77; 85) points, in the group of stimulating spirometry – 81 (72; 85) points in the group with positive expiratory pressure – 79 (73; 84) points (χ2 = 4.560; p = 0.102). Conclusion. Additional respiratory physical therapy did not affect the overall satisfaction rate and results of most items of the questionnaire. The differences found in the three items did not indicate any effect of the use of respiratory physical therapy on the level of satisfaction, as higher scores in these items were in the control group. All items in the questionnaire received high scores

Transfer Learning in Collaborative Filtering with Uncertain Ratings

To solve the sparsity problem in collaborative filtering, researchers have introduced transfer learning as a viable approach to make use of auxiliary data. Most previous transfer learning works in collaborative filtering have focused on exploiting point-wise ratings such as numerical ratings, stars, or binary ratings of likes/dislikes. However, in many real-world recommender systems, many users may be unwilling or unlikely to rate items with precision.In contrast, practitioners can turn to various non-preference data to estimate a range or rating distribution of a user's preference on an item.Such a range or rating distribution is called an uncertain rating since it represents a rating spectrum of uncertainty instead of an accurate point-wise score. In this paper, we propose an efficient transfer learning solution for collaborative filtering, known as {\em transfer by integrative factorization} (TIF), to leverage such auxiliary uncertain ratings to improve the performance of recommendation. In particular, we integrate auxiliary data of uncertain ratings as additional constraints in the target matrix factorization problem, and learn an expected rating value for each uncertain rating automatically. The advantages of our proposed approach include the efficiency and the improved effectiveness of collaborative filtering, showing that incorporating the auxiliary data of uncertain ratings can really bring a benefit. Experimental results on two movie recommendation tasks show that our TIF algorithm performs significantly better over a state-of-the-art non-transfer learning method.

Nearest neighbors algorithm and genetic‐based collaborative filtering

AbstractConventional recommender systems often utilize similarity formulas to identify similarities between active users and others to predict the rating of the unseen items. Existing optimization algorithms seek to find the weights and coefficients affecting these similarities. Our proposed method, implemented in R in the GACFF package, shifts away from this view and directly uses the continuous genetic algorithm to find optimal similarities in big data (e.g., Movielens 1M and Netflix datasets) to improve the performance of user‐based collaborative filtering recommendation systems. First, by identifying the users who are the nearest neighbors along with their number, the number of genes in a chromosome is determined. Each gene represents the similarity between a neighboring user and an active user. This genetic algorithm is independent of the size of the data. Our method provides optimal solutions more quickly by estimating the starting points. Moreover, the genetic metric provides better results and recommendations than previous ones in terms of runtime and quality measures (i.e., mean absolute error, coverage, precision, and recall).

Limited impact of a top-down approach to improve enhanced recovery programme in French university hospitals: a before-after retrospective survey

BackgroundEnhanced recovery programme (ERP) after surgery needs development in Assistance Publique Hôpitaux de Paris (APHP).MethodsA retrospective before-and-after study was performed in 2015 and 2017 on three surgical models (total knee arthroplasty (TKA), colectomy and hysterectomy) in 17 hospitals including 29 surgical departments. Data were collected in one control intervention (total hip arthroplasty (THA), gastrectomy and ovariectomy). In 2016, Massive Open Online Course on ERP and a day meeting information were developed by APHP. A national update on ERP was also organized by HAS and a regional professional partnership programme was started. Primary outcomes were length of stay (LOS) and complications after surgery. Data on ERP items were collected in the patients’ chart and in anaesthetist and surgeon interview. Seventy percent application rate reflects application of ERP procedure.Results1321 patient’s files were analysed (812 in 2015 and 509 in 2017). The LOS (mean (SD)) is reduced by 1.6 day for TKA (2015, 8.7 (6.7) versus 7.1 (3.4) in 2017; p<0.001) but stable for colectomy and hysterectomy. Incidence of severe complications after surgery is unchanged in all types of surgical models. For TKA and hysterectomy respectively applied items of ERP (i.e. >70% application) increased respectively from 5 to 7 out of 17 and 16 in 2015 and 2017. For colectomy, they were stable at 6 out of 21 in 2015 and 2017. The mean application rates of ERP items stayed below 50% in all cases in 2017. The LOS was negatively correlated with ERP items’ application when data collected in 2015 and 2017 were analysed together.ConclusionERP application did not significantly improved between 2015 and 2017 for three surgical models after an institutional information and diffusion of recommendations in 29 surgical departments of seventeen French University hospitals underlining the limit of a top-down approach.

Simulating real profiles for shilling attacks: A generative approach

Collaborative Filtering (CF) approaches are vulnerable to Shilling Attacks, in which malicious users or companies inject a large number of fake profiles in a system in order to manipulate its recommendations. One problem of current Shilling Attack models is that they commonly use straightforward statistical templates, producing profiles with different rating patterns than actual system data, which facilitates its detection, requiring a larger amount of profiles to achieve its goals. To address this problem and create profiles closer to reality, we propose using a generative model, Variational Autoencoder (VAE) to map original data distribution. With VAE, it is possible to generate new profiles based on real data, without explicit copying their actual ratings. Its generated profiles are converted to malicious profiles by adding target item rating value. We test our attack model on MovieLens 100k data set and compare to literature attack models. Our results indicate that our model outperforms all other models in model-based CF system, especially using low attack sizes (from 3% to 5%). Also, analysis comparing profiles generated from it and other approaches shows that our model ratings pattern are very similar to real profiles, which may indicate that attacks mounted using our approach may be less likely to be detected by detection approaches. Thus, we show that our attack model represents an advance on Shilling Attack models, since its superior results in model-based CF and possible indistinction from real profiles may be useful as a baseline to test detection techniques and other tasks among Shilling Attack area.