36-item Form Research Articles

Students with Aberrant Responses in TIMSS2019 Mathematics Test in Sultanate of Oman: Comparison of Between Squared Residual Index by Content, Cognitive Process, Item Type, and Difficulty.

This study aimed to detect the students with misfit responses for eight grade students in the Sultanate of Oman in 2019, who amounting 6,745 students, by using the Between Squared Residual Index over sections of TIMSS2019 Mathematics test (content, cognitive process, item format and item difficulty). This study also aimed to compare the results of the Between Squared Residual Index over Mathematics test sections with the results of the Squared Residual Index over item level. The results of the study indicated that the percentage of students with misfit responses ranged between (2.16% & 3.22%) and between (2.13% & 3.88%) according to the (unweighted and weighted) Between Squared Residual indices respectively, while their percentage amounted to only 0.30% & 0.95% according to the (unweighted & weighed) Squared Residual indices respectively. The results showed that most of students' abilities with misfit responses in the Mathematics Test in TIMSS2019 ranged between low and medium. The study recommended conducting studies to determine the reasons for the presence of misfit responses in Mathematics tests in TIMSS among students of the Sultanate. It also recommended directing teachers and those in charge of applying TIMSS tests to train students in the basic mathematical skills they need to reduce behaviors that lead to misfit responses.

Case Characterization and Perceptions of Athletic Trainers Regarding Medical Disqualification Following Concussion.

Medical disqualification (MDQ) following concussion is a challenging decision clinicians may encounter with little evidence-based guidance. We aimed to 1) describe the MDQ following concussion cases athletic trainers (ATs) have been involved in, 2) describe beliefs about MDQ following concussion, and 3) explore factors that ATs believed should be involved in the MDQ following concussion process. Mixed methods. Online cross-sectional survey with follow-up semi-structured interviews. ATs (n=502) employed at the collegiate setting completed a survey (completion rate=82.3%, n=413/502; male=175, 34.9%; female=235, 46.8%, prefer not to answer=4, 0.8%; no response=88, 17.5%; age=35.3±10.8 years). Twenty participants were also interviewed (males=13, 65.0%; females=7, 35.0%; average age=40.7±11.0years). Participants completed a cross-sectional survey comprised of three sections of MDQ experience and specific case information, MDQ beliefs, and demographic items. We also interviewed participants that completed the survey and indicated involvement in at least one MDQ following concussion case. We addressed aims 1 and 2 using descriptive statistics and aim 3 with a five-cycle content analysis. Nearly half of respondents had been involved in an MDQ case following concussion (49.0% n=246; not involved=51.0%, n=256). ATs who had been involved in at least one MDQ case had involvement in an average of 2.3±1.9 cases (n=241). Participants often described many factors they believed should influence the MDQ decision including sport type, concussion history and recovery, health-related quality of life, and academic performance. Our findings highlight that nearly half of participants were involved in an MDQ case following concussion and navigated this process without guidelines. Given this, multiple factors were considered to evaluate the patient's well-being holistically. The number of ATs involved in MDQ cases following concussion and factors that guided this process warrant further research to develop evidence-based recommendations that assist clinicians in these difficult decisions.

Dietary pattern and diversity analysis using DietDiveR in R: a cross-sectional evaluation in the National Health and Nutrition Examination Survey

BackgroundThere are few resources available for researchers aiming to conduct 24-h dietary record and recall analysis using R. ObjectivesWe aimed to develop DietDiveR, which is a toolkit of functions written in R for the analysis of recall or record data collected with the Automated Self-Administered 24-h Dietary Assessment Tool or 2-d 24-h dietary recall data from the National Health and Nutrition Examination Survey (NHANES). The R functions are intended for food and nutrition researchers who are not computational experts. MethodsDietDiveR provides users with functions to 1) clean dietary data, 2) analyze 24-h dietary intakes in relation to other study-specific metadata variables, 3) visualize percentages of energy intake from macronutrients, 4) perform principal component analysis or k-means clustering to group participants by similar data-driven dietary patterns, 5) generate foodtrees based on the hierarchical food group information for food items consumed, 6) perform principal coordinate analysis taking food grouping information into account, and 7) calculate diversity metrics for overall diet and specific food groups. DietDiveR includes a self-paced tutorial on a website (https://computational-nutrition-lab.github.io/DietDiveR/). As a demonstration, we applied DietDiveR to a demonstration data set and data from NHANES 2015–2016 to derive a dietary diversity measure of nuts, seeds, and legumes consumption. ResultsAdult participants in the NHANES 2015–2016 cycle were grouped depending on the diversity in their mean consumption of nuts, seeds, and legumes. The group with the highest diversity in nuts, seeds, and legumes consumption had 3.8 cm lower waist circumference (95% confidence interval: 1.0, 6.5) than those who did not consume nuts, seeds, and legumes. ConclusionsDietDiveR enables users to visualize dietary data and conduct data-driven dietary pattern analyses using R to answer research questions regarding diet. As a demonstration of this toolkit, we explored the diversity of nuts, seeds, and legumes consumption to highlight some of the ways DietDiveR can be used for analyses of dietary diversity.

Meniscal Comma Sign Responds to Partial Meniscectomy Despite Increased Levels of Arthritis

PurposeTo compare the outcomes of patients undergoing partial meniscectomy preoperatively identified with the “meniscal comma sign” with those undergoing meniscectomy with other tear patterns. MethodsPatients with meniscal “comma sign,” as indicated by a query of MRI reports, were screened using the search terms “meniscotibial recess,” “meniscus perched over the medial tibial margin,” or other search terms by radiologists between January 2008 and November 2019. Patients were matched and chart review was done for demographics, revision surgery, and progression to total knee arthroplasty. Radiographs were used for osteoarthritis grading using the Kellgren-Lawrence (KL) scoring system. Preoperative and postoperative International Knee Documentation Committee (IKDC), Knee Injury and Osteoarthritis Outcome Score (KOOS), Lysholm, and Short Form 12-item Survey (SF-12) scores were collected. Results406 patients met inclusion (comma sign =197; control group n=209). The control group had an increased duration of symptoms at the initial visit (P=0.001). More meniscal comma sign patients received corticosteroid knee injections before surgery (P=0.011). Comma sign patients also had greater mean KL scores (P=0.001) as well as higher KL categorical scores (P=0.002), indicating more advanced levels or arthritis. There were no differences in those receiving physical therapy (PT) before surgery (P=0.966) or those receiving injections or PT after surgery (P=0.631, P=0.37, respectively). IKDC, KOOS, Lysholm, and SF-12 Physical scores improved preoperatively to postoperatively in both groups (P<0.05), and there was no difference between the case and control group (P>0.05). No significant difference was found in revisions or progression to total knee arthroplasty between cohorts. (P=0.676 and P=0.424). ConclusionPatients presenting with preoperative findings of meniscal comma sign fare similarly to those that do not. Patients with this meniscal injury tend to have more advanced grading of osteoarthritic changes in the knee upon presentation and seek care earlier than those without. Arthroscopic meniscectomy is a good treatment option for patients with a meniscal fragment in the meniscotibial recess and shows outcomes comparable to those with other tear patterns. Level of EvidenceLevel III, retrospective cohort

Dual-view multi-modal contrastive learning for graph-based recommender systems

Personalized recommender systems play a crucial role in various online content-sharing platforms (e.g., TikTok). The learning of representations for multi-modal content is pivotal in current graph-based recommender systems. Existing works aim to enhance recommendation accuracy by leveraging multi-modal features (e.g., image, sound, text) as side information for items. However, this approach falls short in fully discerning users' fine-grained preferences across different modalities. To tackle this limitation, this paper introduces the Dual-view Multi-Modal contrastive learning Recommendation model (DMM-Rec). DMM-Rec employs self-supervised learning to guide the learning of user and item representations within the multi-modal context. Specifically, to capture users' preferences for different modalities, we propose specific-modal contrastive learning. Simultaneously, to capture users' cross-modal preferences, cross-modal contrastive learning is introduced to uncover interdependencies in users' preferences across modalities. The contrastive learning tasks not only adaptively explore potential relations between modalities but also address the data sparsity challenge in recommender systems. Extensive experiments conducted on three datasets and compared against ten baselines demonstrate that DMM-Rec outperforms the strongest baseline by an average of 6.81%. These results underscore the effectiveness of considering multi-modal content in improving recommender systems.

Development of a rapid tool for screening financial toxicity based on the comprehensive score for financial toxicity

ObjectiveThis study aimed to use item response theory (IRT) to develop and validate a short version of the Chinese COmprehensive Score for financial Toxicity (COST-S-C) in patients with cancer. MethodsA total of 500 patients with cancer participated in the survey. Confirmatory factor analysis (CFA) was used to assess the factor structure of the COST-S-C. A two-parameter IRT model was used to examine the item fit of the COST-S-C. Item discrimination, item characteristic curves (ICCs), item information curves (IICs), and item fit were estimated. Differential item functioning (DIF) was assessed for participants’ gender and age. Reliability, convergent and discriminant validity was examined as well as estimating the optimal cut-off points of the COST-S-C. ResultsThe CFA results supported the bi-factor structure of the COST-S-C. The IRT analysis findings revealed that several items showed problems with the ICCs, providing little information in terms of IICs as well as poor discrimination. Item 5 showed a negligible DIF problem with age. A valid 3-item COST-S-C scale and its optimal cut-off point was developed. ConclusionsThe COST-S-C is a valid and quick screening tool used to distinguish patients with cancer who report significant financial toxicity in various medical settings.

LLMRG: Improving Recommendations through Large Language Model Reasoning Graphs

Recommendation systems aim to provide users with relevant suggestions, but often lack interpretability and fail to capture higher-level semantic relationships between user behaviors and profiles. In this paper, we propose a novel approach that leverages large language models (LLMs) to construct personalized reasoning graphs. These graphs link a user's profile and behavioral sequences through causal and logical inferences, representing the user's interests in an interpretable way. Our approach, LLM reasoning graphs (LLMRG), has four components: chained graph reasoning, divergent extension, self-verification and scoring, and knowledge base self-improvement. The resulting reasoning graph is encoded using graph neural networks, which serves as additional input to improve conventional recommender systems, without requiring extra user or item information. Our approach demonstrates how LLMs can enable more logical and interpretable recommender systems through personalized reasoning graphs. LLMRG allows recommendations to benefit from both engineered recommendation systems and LLM-derived reasoning graphs. We demonstrate the effectiveness of LLMRG on benchmarks and real-world scenarios in enhancing base recommendation models.

Graph Bayesian Optimization for Multiplex Influence Maximization

Influence maximization (IM) is the problem of identifying a limited number of initial influential users within a social network to maximize the number of influenced users. However, previous research has mostly focused on individual information propagation, neglecting the simultaneous and interactive dissemination of multiple information items. In reality, when users encounter a piece of information, such as a smartphone product, they often associate it with related products in their minds, such as earphones or computers from the same brand. Additionally, information platforms frequently recommend related content to users, amplifying this cascading effect and leading to multiplex influence diffusion. This paper first formulates the Multiplex Influence Maximization (Multi-IM) problem using multiplex diffusion models with an information association mechanism. In this problem, the seed set is a combination of influential users and information. To effectively manage the combinatorial complexity, we propose Graph Bayesian Optimization for Multi-IM (GBIM). The multiplex diffusion process is thoroughly investigated using a highly effective global kernelized attention message-passing module. This module, in conjunction with Bayesian linear regression (BLR), produces a scalable surrogate model. A data acquisition module incorporating the exploration-exploitation trade-off is developed to optimize the seed set further. Extensive experiments on synthetic and real-world datasets have proven our proposed framework effective. The code is available at https://github.com/zirui-yuan/GBIM.

Reviewing the Design Features of the Microstructures of Two Popular Chinese-English Learners’ Dictionaries Published in China

Abstract Based on the design feature framework of Chinese–English (C–E) learners’ dictionaries (with some modifications made) proposed by Zhang and Hu (2023), this study aims to review the design features of the microstructures of two popular C–E learners’ dictionaries published in China. The results show: 1) the two dictionaries do not distinguish between Chinese and foreign users, nor do they specify target users’ language proficiency; 2) some information items and their contents are not appropriately included and treated, even contrary to the primary function of the two dictionaries; 3) the ordering principles of senses, English equivalents and examples are not pertinent to the primary function of the two dictionaries, or even without any principles at all. The cause of the problems in the two dictionaries lies in the neglect of the study and application of systematic design feature frameworks. Therefore, in compiling C–E learners’ dictionaries, lexicographers should strengthen the study and application of systematic design feature frameworks.

Enhancing Retailer Satisfaction: Strategies and Best Practices

Within the energetic and competitive scene of the retail industry, the fulfillment of retailers plays a essential part in deciding the victory and supportability of supplier-retailer associations. This long theoretical investigates comprehensive methodologies and best hones pointed at upgrading retailer fulfillment, in this manner cultivating more grounded and more affluent connections between providers and retailers. Understanding the different needs and inclinations of retailers is foundational to any exertion pointed at upgrading fulfillment. Providers ought to utilize different strategies, counting standard overviews, input components, and advertise examination, to pick up bits of knowledge into retailer desires, requests, and showcase patterns. By understanding these components, providers can tailor their offerings and administrations to meet the particular prerequisites of retailers, hence upgrading fulfillment. Consistency in item quality is vital in keeping up retailer fulfillment. Providers must actualize strong quality control measures all through the generation and supply chain to guarantee that items reliably meet or surpass the desires of retailers and their clients. Clear details, rules for item dealing with and capacity, and responsive back for quality-related issues contribute to building believe and certainty among retailers. Proficiency in requesting and conveyance forms is another basic angle of upgrading retailer fulfillment. Providers ought to contribute in user-friendly online requesting stages, streamlined coordinations, and adaptable conveyance choices to disentangle the obtainment prepare for retailers. Opportune and exact conveyances are fundamental for minimizing disturbances to retailers’ operations and keeping up believe and unwavering quality. Remarkable client bolster is irreplaceable in tending to retailers’ request, concerns, and issues expeditiously and viably. Providers ought to keep up devoted back channels staffed with learned and responsive agents. Standard preparing sessions can enable bolster staff with item information and communication aptitudes, empowering them to supply personalized help and investigating direction to retailers. Building solid and collaborative connections is crucial to cultivating retailer fulfillment. Providers ought to develop a organization mentality based on believe, straightforwardness, and shared regard. Organizing occasions, workshops, and motivation programs can encourage interaction and collaboration between providers and retailers, cultivating a sense of community and shared objectives Nonstop advancement and advancement are key to remaining ahead within the competitive retail scene. Providers ought to request progressing input from retailers to distinguish zones for enhancement and contribute in investigate and improvement to present modern items, highlights, and administrations custom fitted to retailer needs and inclinations. Checking and measuring fulfillment measurements are basic for assessing execution and identifying areas for improvement. Key execution markers such as arrange precision, on-time conveyance, and client criticism scores can give profitable bits of knowledge into the adequacy of supplier-retailer intuitive. Leveraging information analytics devices empowers providers to pick up more profound experiences into retailer behavior and inclinations, encouraging educated decision-making and methodology refinement. In conclusion, upgrading retailer fulfillment requires a multifaceted approach that envelops understanding retailer needs, guaranteeing item quality and unwavering quality, streamlining forms, giving remarkable bolster, building solid connections, cultivating innovation, and checking execution. By prioritizing these methodologies and best hones, providers can develop more grounded associations with retailers, driving common victory and competitive advantage within the retail industry.

Robust social recommendation based on contrastive learning and dual-stage graph neural network

GNN-based social recommendation aims to use social network information to improve recommendation performance of traditional user–item interaction network (U–I network). However, in graph neural network information aggregation, both social networks and U–I networks inevitably have noise, which affects accuracy of recommendation results. To reduce the noise impact of network data, we propose Robust Social Recommendation based on Contrastive Learning and Dual-Stage Graph Neural Network (CLDS). First, considering instability of social networks, we propose the social preference network. It is robust and retains only social friend relationships with common preferences. Based on it and U–I network, we construct a social recommendation pre-training model. Next, we propose self-contrastive learning method. The method initializes multiple social network node representations through Gaussian distribution, pre-training and random disturbance, respectively. Then, it uses contrastive learning on the generated multiple node representations to enhance the robustness of node representation. Finally, CLDS avoids directly capturing potentially user–user and item–item information in U–I networks which is incomplete and untrusted. And instead, it only extracts user–item information to reduce the noise generated by GNN-based U–I network information aggregation. We conduct experiments under the open-source real network dataset. The experimental results show that CLDS outperforms state-of-art methods in social recommendation. The code is available at: https://github.com/Andrewsama/CLDS-master.

A graph neural network with topic relation heterogeneous multi-level cross-item information for session-based recommendation

The aim of session-based recommendation (SBR) mainly analyzes the anonymous user’s historical behavior records to predict the next possible interaction item and recommend the result to the user. However, due to the anonymity of users and the sparsity of behavior records, recommendation results are often inaccurate. The existing SBR models mainly consider the order of items within a session and rarely analyze the complex transition relationship between items, and additionally, they are inadequate at mining higher-order hidden relationship between different sessions. To address these issues, we propose a topic relation heterogeneous multi-level cross-item information graph neural network (TRHMCI-GNN) to improve the performance of recommendation. The model attempts to capture hidden relationship between items through topic classification and build a topic relation heterogeneous cross-item global graph. The graph contains inter-session cross-item information as well as hidden topic relation among sessions. In addition, a self-loop star graph is established to learn the intra-session cross-item information, and the self-connection attributes are added to fuse the information of each item itself. By using channel-hybrid attention mechanism, the item information of different levels is pooled by two channels: max-pooling and mean-pooling, which effectively fuse the item information of cross-item global graph and self-loop star graph. In this way, the model captures the global information of the target item and its individual features, and the label smoothing operation is added for recommendation. Extensive experimental results demonstrate that the recommendation performance of TRHMCI-GNN model is superior to the comparable baseline models on the three real datasets Diginetica, Yoochoose1/64 and Tmall. The code is available now.11https://github.com/usstyangfan/TRHMCI-GNN.

Outcomes of Image-Free Robotic Assisted Total Knee Arthroplasty in Patients Who Have Valgus Knee Deformities

BackgroundValgus knee deformities pose a unique challenge in total knee arthroplasty (TKA) due to the complexity of achieving ligamentous balance and satisfactory alignment compared to varus or neutral deformities. Robotic-assisted (RA) TKA could aid in achieving improved component alignment and balance. MethodsWe retrospectively evaluated a matched cohort of patients to compare image-free RA-TKA (n = 44) versus conventional manual (CM) TKA (n = 30) techniques in patients who have valgus deformity of 5 to 15 degrees, including radiographic and patient-reported outcomes measures (PROMs) over a 3-year period. The patient reported outcome measures (PROMs) studied to determine outcomes were: Western Ontario McMaster University Arthritis Index, Knee Society Score-Function Score, and Short Form 12-item Survey. ResultsOverall, the RA-TKA cohort showed faster improvement in PROMs (37.16 ± 1 8.8 versus 25.74 ± 17.7, P = .02), shorter length of stay (1.41 versus 2.29 days, P = .02), and shorter operating room times (120.79 versus 123.67 minutes, P = .02) than CM-TK). Additionally, there was no difference in the use of primary versus varus-valgus constrained polyethylene liners between the cohorts. ConclusionsIn this investigation, RA-TKA yielded a slightly faster patient recovery, more objective measurements of ligamentous balance, and proved noninferior PROMs compared to CM-TKA for preoperative valgus knee deformities.

Recommendation Approach Based on TwinBERT Model

In order to provide more personalized recommendation services for users as well as improve the platform's traffic and user satisfaction, a recommendation method based on the TwinBERT model has been proposed. As a matter of fact, this method combines not only the basic information of users and items but also user comments or item text information to predict users' behaviors and make more accurate recommendations. The text information in the dataset is encoded in sentences using the BERT model, and the encoded text features are fused with other features to obtain a representation of user and item embedding vectors. These vectors are then used to predict the user's rating of the item and recommend relevant items for the user. The proposed method was tested on the Amazon review dataset and the MovieLens dataset, and the experimental results showed that it improved over traditional recommendation algorithms in terms of accuracy and efficiency.

A Proposal of a Novel Automatic Checkout System Reducing Additional Item Information Relearning Cost

Recently, a retail industry has trended to adopt automatic checkout systems for enhancing customer sales. Typically, an automatic checkout system employs an object detection technique (i.e., image processing) based on a commonly used deep learning model (e.g., Faster R‐CNN and YOLO) that has been trained for items. Therefore, it takes much time to relearn the model whenever an item is added to the system. To save the relearning cost, we propose a novel automatic checkout system reducing the relearning cost of an additional item. © 2024 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Item-Explanatory Mokken Scale Analysis: Using Nonparametric Item Response Theory to Explore Item Attributes

Researchers frequently examine the effects of item attributes (i.e., item properties, item covariates, item formats) as part of item analysis for cognitive and affective assessments. In this study, we propose and illustrate item-explanatory Mokken Scale Analysis (MSA) as an exploratory nonparametric approach to examining item attributes based on invariance principles. We used real data and an illustrative simulation to demonstrate the approach under various conditions. Our results suggest that item-explanatory MSA provides practical and theoretical insight to support measurement research in a variety of contexts. Because of the emphasis on invariance, our approach can help researchers explore the impact of item characteristics on examinee responses and evaluate the degree to which item characteristics have a consistent interpretation for all examinees. We consider the implications of our findings for research and practice, along with directions for further investigation.

Enhancing Patient-Reported Outcome Measures for Drug Hypersensitivity: A Validated 6-Item Quality-of-Life Questionnaire for Patients With Drug Hypersensitivity

BACKGROUNDDrug hypersensitivity reactions (DHR) can significantly impair patients’ health-related quality of life (HRQoL). However, tools for HRQoL assessment for patients with DHR are time-consuming and remain underutilized. OBJECTIVETo develop and validate an optimized version of the Drug Hypersensitivity Quality of Life Questionnaire (DrHy-Q) designed for everyday clinical use. METHODSItem response theory (IRT), a statistical framework for psychometric measurement, was used to evaluate the 15 questions from the original DrHy-Q for their respective item difficulty, discrimination, and information using prospective data from 243 patients with histories of suspected/confirmed DHR before allergy workup. Accordingly, the best-performing items were identified to develop a 6-item optimized version (DrHy-Q6), which was subsequently validated with another prospective cohort of 156 patients. RESULTSAll 15 items of the original DrHy-Q demonstrated satisfactory parameters in IRT analysis, including very high discrimination (>1.7), appropriate difficulty (in between -1.5 and 1.5) and good information (a high and broad peak in the information curve). Six items with top-ranked IRT parameters were identified to construct an optimized version, which we named the DrHy-Q6. The DrHy-Q6 demonstrated a 1-factor structure with an improved fit compared to the original DrHy-Q (comparative fit index = 0.985, Tucker–Lewis’s index = 0.974), excellent convergent validity (Unadjusted Pearson correlation with the full version = 0.955; adjusted = 0.894, p<0.001), reliability (Cronbach’s α and McDonald’s ω = 0.93), divergent validity (Pearson correlation with all Short Form 12-item Healthy Survey Version 2 [SF-12v2] subscales < 0.60, p<0.001) and discriminant validity (significantly higher scores with multiple DHR labels [42.45 ± 27.26 vs. 32.93 ± 26.66], p=0.013). CONCLUSIONSFrom an IRT perspective, the DrHy-Q and all its constituent items are psychometrically valid for HRQoL assessment. We propose an optimized 6-item version (DrHy-Q6) as an abbreviated alternative for assessing HRQoL in patients with DHR, especially for routine use in clinical practice. Patients and physicians may benefit from its streamlined length and simpler scoring algorithm.

Self-supervised hypergraph neural network for session-based recommendation supported by user continuous topic intent

Session-based recommendation aims to predict the next item an anonymous user is most likely to click based on its limited historical behaviors. However, in session-based recommendation, the anonymity and short sequences often lead to the issue of data sparsity, which can affect the performance of recommendation models. Existing recommendation methods mostly focus on individual items in the session sequence, without considering the implicit user continuous intent underlying multiple continuous items. Moreover, current graph-based recommendation models do not fully take into account the direction of information transfer among shared items when calculating their similarities. This paper proposes, TCAUIS, a topic classification augmented user intend self-supervised hypergraph neural network for session-based recommendation, which contains continuous topic intent of users within a session and the information of co-occurrence items. The TCAUIS models a individual session sequence as a topic intent hypergraph and co-occurrence graph respectively, and employs latent Dirichlet allocation to classify the topic of the dataset and fuses the topic hyperedges into the hypergraph to alleviate the sparsity of the data and enhance the session information. Additionally, the hypergraph attention networks and attention mechanisms are leveraged to capture the complex high-order information between nodes in both the hypergraphs and co-occurrence graphs. The continuous topic intent of users and information of co-occurrence relationship are effectively integrated using a self-supervised mechanism that maximizes the mutual information between the representations of two sessions. Extensive experimental results show that the proposed TCAUIS model outperforms the state-of-the-art models on three real datasets Diginetica, Yoochoose1/64 and Nowplaying.

Hybrid Prompt Recommendation Explanation Generation combined with Graph Encoder

Recommendation systems have been effectively utilized in various fields, but their internal decision-making methods are still largely unknown. This opaque decision-making method can greatly affect users’ trust in the recommendation system. Therefore, finding a way to explain the reasons for model decisions has become an urgent task. Previous studies often used LSTM and other models to generate recommendation explanations and explain the reasons for recommendations in text form. However, traditional methods cannot effectively use the ID information of users and items, and the text generated is highly repetitive. To solve this problem, this paper uses the method of prompt learning combined with a graph encoder to design a recommendation explanation generation model. In order to narrow the semantic gap between the ID information of users and items and natural language and capture high-level interaction information, this paper designs a graph encoder based on user similarity to learn the interactive semantic information of user and item IDs, and to construct a continuous prompt. Then, the discrete prompt composed of discrete features of users and items is combined with the continuous prompt to construct a hybrid prompt to input into the pre-trained model to generate the recommended explanation. This paper experiments on three publicly available datasets and compares them with several state-of-the-art methods to demonstrate the personalization and text quality of the generated explanations.

Set representative vector and its asymmetric attention-based transformation for heterogeneous set-to-set matching

Heterogeneous set-to-set matching, applied to fashion outfit recommendations, no longer depends on the similarity but on compatibility between items in sets. Existing state-of-the-art methods apply self- and cross-attention mechanisms to transform item vectors closer or farther apart and compute the matching score based on transformed item vectors. However, the transformation by attention mechanisms with fewer items in sets is performed in rather limited spaces, and the complex computation of scores in the head network would cause the instability of the training of entire networks. To overcome these problems, we propose to add a trainable set representative vector into each set and to embed discriminative information of items onto the vector through originally extended asymmetric attention mechanisms. This enables dynamic transformation in wider spaces and stable training without gradient vanishing problems to generate discriminative matching scores. Through experiments with heterogeneous set-to-set matching tasks, even-total, and fashion outfit matching tasks, we show the effectiveness of our proposed method.