Dataset Information

Deep transfer learning with multimodal embedding to tackle cold-start and sparsity issues in recommendation system.

ABSTRACT: Recommender systems (RSs) have become increasingly vital in the modern information era and connected economy. They play a key role in business operations by generating personalized suggestions and minimizing information overload. However, the performance of traditional RSs is limited by data sparseness and cold-start issues. Though deep learning-based recommender systems (DLRSs) are very popular, they underperform when considering rating matrices with sparse entries. Despite their performance improvements, DLRSs also suffer from data sparsity, cold start, serendipity, and generalizability issues. We propose a multistage model that uses multimodal data embedding and deep transfer learning for effective and personalized product recommendations, and is designed to overcome data sparsity and cold-start issues. The proposed model includes two phases. In the first-offline-phase, a deep learning technique is implemented to learn hidden features from a large image dataset (targeting new item cold start), and a multimodal data embedding is used to produce dense user feature and item feature vectors (targeting user cold start). This phase produces three different similarity matrices that are used as inputs for the second-online-phase to generate a list of top-n relevant items for a target user. We analyzed the accuracy and effectiveness of the proposed model against the existing baseline RSs using a Brazilian E-commerce dataset. The results show that our model scored 0.5882 for MAE and 0.4011 for RMSE which is lower than baseline RSs which indicates that the model achieved an improved accuracy and was able to minimize the typical cold start and data sparseness issues during the recommendation process.

SUBMITTER: Jafri SIH

PROVIDER: S-EPMC9410545 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep transfer learning with multimodal embedding to tackle cold-start and sparsity issues in recommendation system.

Jafri Syed Irteza Hussain SIH Ghazali Rozaida R Javid Irfan I Mahmood Zahid Z Hassan Abdullahi Abdi Abubakar AAA

PloS one 20220825 8

Recommender systems (RSs) have become increasingly vital in the modern information era and connected economy. They play a key role in business operations by generating personalized suggestions and minimizing information overload. However, the performance of traditional RSs is limited by data sparseness and cold-start issues. Though deep learning-based recommender systems (DLRSs) are very popular, they underperform when considering rating matrices with sparse entries. Despite their performance im ...[more]

PMID: 36007091

Similar Datasets

Project description:ContextAmerican football is associated with the largest number of emergency department visits for pediatric sport-related traumatic brain injury, including concussions. Tackling is responsible for almost two-thirds of football concussions. Some have recommended implementing age restrictions on tackling in youth football.ObjectiveTo determine whether the public would support age restrictions and the factors that may drive such support.DesignCross-sectional study.SettingSummer wave of Porter Novelli's 2020 US ConsumerStyles survey.ParticipantsA total of 4053 adults.Main outcome measure(s)Respondents answered questions about a good age for children to start playing tackle football, past football playing history, concerns about safety, and the benefits and risks of playing football.ResultsMost respondents believed that middle (32.8%) or high (27.8%) school was a good age to start playing tackle football. About 1 in 5 (19.8%) respondents reported that children should never play tackle football. Certain groups of individuals were more likely to support children starting to play tackle football at high school age or above or to say that they should never play tackle football, including those with a bachelor's degree or higher (rate ratio [RR]HS+ = 1.41, 95% CI = 1.14, 1.76; RRNEVER = 2.70, 95% CI = 1.93, 3.78), those who did not have children under 18 (RRHS+ = 1.54, 95% CI = 1.26, 1.90; RRNEVER = 1.54, 95% CI = 1.14, 2.07), those who were not football fans (RRNEVER = 3.07, 95% CI = 2.32, 4.06), and those who were very (RRHS+ = 3.94, 95% CI = 2.87, 5.42; RRNEVER = 11.52, 95% CI = 7.32, 18.15) or somewhat (RRHS+ = 1.88, 95% CI = 1.41, 2.52) concerned about the safety of children.ConclusionDespite acknowledging the benefits of playing football, many adults expressed concern about safety and endorsed high school age and older or never as a good age to start playing tackle football, highlighting a disconnect with current football program practices regarding age.

Project description:BackgroundThe multiplicity of issues, including uncertainty and ethical dilemmas, and policies involved in appraising interventions for rare diseases suggests that multicriteria decision analysis (MCDA) based on a holistic definition of value is uniquely suited for this purpose. The objective of this study was to analyze and further develop a comprehensive MCDA framework (EVIDEM) to address rare disease issues and policies, while maintaining its applicability across disease areas.MethodsSpecific issues and policies for rare diseases were identified through literature review. Ethical and methodological foundations of the EVIDEM framework v3.0 were systematically analyzed from the perspective of these issues, and policies and modifications of the framework were performed accordingly to ensure their integration.ResultsAnalysis showed that the framework integrates ethical dilemmas and issues inherent to appraising interventions for rare diseases but required further integration of specific aspects. Modification thus included the addition of subcriteria to further differentiate disease severity, disease-specific treatment outcomes, and economic consequences of interventions for rare diseases. Scoring scales were further developed to include negative scales for all comparative criteria. A methodology was established to incorporate context-specific population priorities and policies, such as those for rare diseases, into the quantitative part of the framework. This design allows making more explicit trade-offs between competing ethical positions of fairness (prioritization of those who are worst off), the goal of benefiting as many people as possible, the imperative to help, and wise use of knowledge and resources. It also allows addressing variability in institutional policies regarding prioritization of specific disease areas, in addition to existing uncertainty analysis available from EVIDEM.ConclusionThe adapted framework measures value in its widest sense, while being responsive to rare disease issues and policies. It provides an operationalizable platform to integrate values, competing ethical dilemmas, and uncertainty in appraising healthcare interventions.

Dataset Information

Deep transfer learning with multimodal embedding to tackle cold-start and sparsity issues in recommendation system.

Publications

Deep transfer learning with multimodal embedding to tackle cold-start and sparsity issues in recommendation system.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets