Recipe Recommendation With Hierarchical Graph Attention Network

Recipe recommendation systems play an important role in helping people find recipes that are of their interest and fit their eating habits. Unlike what has been developed for recommending recipes using content-based or collaborative filtering approaches, the relational information among users, recipes, and food items is less explored. In this paper, we leverage the relational information into recipe recommendation and propose a graph learning approach to solve it. In particular, we propose HGAT, a novel hierarchical graph attention network for recipe recommendation. The proposed model can capture user history behavior, recipe content, and relational information through several neural network modules, including type-specific transformation, node-level attention, and relation-level attention. We further introduce a ranking-based objective function to optimize the model. Thorough experiments demonstrate that HGAT outperforms numerous baseline methods.

SessNet: A Hybrid Session-Based Recommender System

<div>Session-based recommendation is the task of predicting user actions during short online sessions. Previous work considers the user to be anonymous in this setting, with no past behavior history available. In reality, this is often not the case, and none of the existing approaches are flexible enough to seamlessly integrate user history when available. In this thesis, we propose a novel hybrid session-based recommender system to perform next-click prediction, which is able to take advantage of historical user preferences when accessible. Specifically, we propose SessNet, a deep profiling session-based recommender system, with a two-stage dichotomy. First, we use bidirectional transformers to model local and global session intent. Second, we concatenate any user information with the current session representation to feed to a feed-forward neural network to identify the next click. Historical user preferences are computed using the sequence-aware embeddings obtained from the first step, allowing us to better understand the users. We evaluate the efficacy of the proposed method using two benchmark datasets, YooChoose1/64 and Dignetica. Our experimental results show that SessNet outperforms state-of-the-art session-based recommenders on P@20 for both datasets.</div>

SessNet: A Hybrid Session-Based Recommender System

<div>Session-based recommendation is the task of predicting user actions during short online sessions. Previous work considers the user to be anonymous in this setting, with no past behavior history available. In reality, this is often not the case, and none of the existing approaches are flexible enough to seamlessly integrate user history when available. In this thesis, we propose a novel hybrid session-based recommender system to perform next-click prediction, which is able to take advantage of historical user preferences when accessible. Specifically, we propose SessNet, a deep profiling session-based recommender system, with a two-stage dichotomy. First, we use bidirectional transformers to model local and global session intent. Second, we concatenate any user information with the current session representation to feed to a feed-forward neural network to identify the next click. Historical user preferences are computed using the sequence-aware embeddings obtained from the first step, allowing us to better understand the users. We evaluate the efficacy of the proposed method using two benchmark datasets, YooChoose1/64 and Dignetica. Our experimental results show that SessNet outperforms state-of-the-art session-based recommenders on P@20 for both datasets.</div>

Movie Recommendation System Mistreatment Current Trends and Sentiment Analysis from Micro Blogging Knowledge

Recommendation systems (RSs) have garnered immense interest for applications in e-commerce and digital media. Traditional approaches in RSs include such as collaborative filtering (CF) and content-based filtering (CBF) through these approaches that have certain limitations, such as the necessity of prior user history and habits for performing the task of recommendation. To minimize the effect of such limitation, this article proposes a hybrid RS for the movies that leverage the best of concepts used from CF and CBF along with sentiment analysis of tweets from microblogging sites. The purpose to use movie tweets is to understand the current trends, public sentiment, and user response of the movie. Experiments conducted on the public database have yielded promising results. Keywords: Collaborative filtering, Content based filtering, Recommendation System, Sentiment Analysis, Twitter

A Survey of Long-Tail Item Recommendation Methods

Recommender systems represent a critical field of AI technology applications. The core function of a recommender system is to recommend items of interest to users, but if it is only user history-based (purchasing or browsing data), it can only recommend similar products to a user, which makes the user feel fatigued (creating so-called “Information Cocoons”). Besides, transaction data (purchasing or browsing data) in various fields usually follow Pareto distributions. Accordingly, 20% of products are purchased or viewed a greater number of times (short-head items), while the remaining 80% of products are purchased or viewed less frequently (long-tail items). Using the traditional recommendation method, considering only the accuracy of recommendations, the coverage rate is relatively low, and most of the recommended items are short-head items. The long-tail item recommendation method not only considers the recommendation of short-head items but also considers recommending more long-tail items to users, thus improving the coverage and diversity of the recommendation results. Long-tail item recommendation research has become a frontier issue in recommendation systems in recent years. While the current research paper is still scarce, there have been related research achievements in top-level conferences in the field of computers, such as VLDB and IJCAI. Due to the fact that there is no review literature in this field, to allow readers to better understand the research status of the long-tail item recommendation method, this paper summarizes the progress of the research on long-tail item recommendation methods (from clustering-based, which began in 2008, to deep learning-based methods, which began in 2020) and the future directions associated with this research.

Deep-Sequence–Aware Candidate Generation for e-Learning System

Recently proposed recommendation systems based on embedding vector technology allow us to utilize a wide range of information such as user side and item side information to predict user preferences. Since there is a lack of ability to use the sequential information of user history, most recommendation system algorithms fail to predict the user’s preferences more accurately. Therefore, in this study, we developed a novel recommendation system that takes advantage of sequence and heterogeneous information in the candidate-generation process. The principle underlying the proposed recommendation model is that the new sequence based embedding layer in the model catches the sequence pattern of user history. The proposed deep-learning model may improve the prediction accuracy using user data, item data, and sequential information of the user’s profile. Experiments were conducted on datasets of the Korean e-learning platform, and the empirical results confirmed the capability of the proposed approach and its superiority over models that do not use the sequences of the heterogeneous information of users and items for the candidate-generation process.

Instagram Post Popularity Trend Analysis and Prediction using Hashtag, Image Assessment, and User History Features

Instagram is one of the most popular social networks for marketing. Predicting the popularity of a post on Instagram is important to determine the influence of a user for marketing purposes. There were studies on popularity prediction on Instagram using various features and datasets. However, they haven't fully addressed the challenge of data variability of the global dataset, where they either used local datasets or discretized output. This research compared several regression techniques to predict the Engagement Rate (ER) of posts using a global dataset. The prediction model, coupled with the results of the popularity trend analysis, will have more utility for a larger audience compared to existing studies. The features were extracted from hashtags, image analysis, and user history. It was found that image quality, posting time, and type of image highly impact ER. The prediction accuracy reached up to 73.1% using the Support Vector Regression (SVR), which is higher than previous studies on a global dataset. User history features were useful in the prediction since the data showed a high variability of ER if compared to a local dataset. The added manual image assessment values were also among the top predictors

Comparative Study of Machine Learning Algorithms for Recommendation System

The role of recommender system is very vital in recent times for a lot of individuals. It helps in taking decisions without exploring physically. Broadly there are two types of recommender system: Content based and Collaborative Filtering. The first one focus on user’s history and takes decisions. But there could be times when decisions based on only user history is not sufficient. For this, there is a need to analyze many parameters influencing the decision such as previous history, Age, gender, location etc. In the second approach it finds similar group of users based on several parameters and then takes decisions. Over the last few decades machine learning algorithms have proved their worth in this area because of their ability to learn from the given data and identify various hidden patterns. With this learning, these algorithms are able to generalize very well for unknown data. In this research work, a survey on three different machine learning based collaborative filtering methods are presented using Movie Lens dataset. The comparison of all three methods based on RMSE and MAE error is also discussed.

An Attention-based Model for Conversion Rate Prediction with Delayed Feedback via Post-click Calibration

Conversion rate (CVR) prediction is becoming increasingly important in the multi-billion dollar online display advertising industry. It has two major challenges: firstly, the scarce user history data is very complicated and non-linear; secondly, the time delay between the clicks and the corresponding conversions can be very large, e.g., ranging from seconds to weeks. Existing models usually suffer from such scarce and delayed conversion behaviors. In this paper, we propose a novel deep learning framework to tackle the two challenges. Specifically, we extract the pre-trained embedding from impressions/clicks to assist in conversion models and propose an inner/self-attention mechanism to capture the fine-grained personalized product purchase interests from the sequential click data. Besides, to overcome the time-delay issue, we calibrate the delay model by learning dynamic hazard function with the abundant post-click data more in line with the real distribution. Empirical experiments with real-world user behavior data prove the effectiveness of the proposed method.