Dataset

This dataset was created by Noor Saeed

Released under Apache 2.0

Contents

Personalized Recommendation Systems Dataset (150,000 Entries)

This dataset is a fictional representation of user interactions within an e-commerce or streaming platform, created specifically for educational and training purposes. It simulates realistic user behavior and interactions to aid in developing and testing machine learning models for personalized recommendation systems. With 150,000 entries, it offers a rich variety of features suitable for building and evaluating algorithms in recommendation systems, user behavior analysis, and predictive modeling.

Dataset Features: 1. User_ID: A unique identifier for each user (e.g., User_1, User_2, etc.), representing individual profiles on the platform.
2. Item_ID: A unique identifier for each item, such as a product, movie, or song.
3. Category: The type of item interacted with (e.g., Electronics, Books, Music, Movies, etc.), providing insights into user preferences.
4. Rating: User-assigned ratings on a scale of 1.0 to 5.0, reflecting the level of satisfaction with the item.
5. Timestamp: The exact date and time of the interaction, useful for time-based analysis.
6. Price: The price of the item at the time of interaction, recorded in USD.
7. Platform: The platform or device used to interact with the system (e.g., Web, Mobile App, Smart TV, Tablet), capturing multi-device behavior.
8. Location: The geographic region of the user, categorized into areas such as North America, Europe, Asia, etc., for regional behavioral analysis.

Applications: This dataset is versatile and can be used for: - Collaborative Filtering Models: Harness user-item interaction data to recommend items based on similar users or items.
- Content-Based Recommendation Systems: Leverage item attributes to generate personalized recommendations.
- User Behavior Analysis: Uncover insights into user preferences, habits, and trends to inform marketing strategies.
- Predictive Modeling: Train machine learning models to predict user preferences or future interactions.

Important Note: This dataset is fictional and does not represent real-world data. It has been generated solely for educational and training purposes, making it ideal for students, researchers, and data scientists who want to practice building machine learning models without using sensitive or proprietary data.

Why Use This Dataset? 1. Diverse and Realistic Features: Simulates key aspects of user interaction in modern platforms.
2. Scalable Size: Provides sufficient data for training advanced machine learning models, ensuring robust validation.
3. Rich Metadata: Enables detailed analysis and multiple use cases, from recommendation systems to business analytics.

This dataset is a great resource for exploring personalized recommendations or enhancing machine learning skills in a practical and safe manner.

Dataset

This dataset was created by Shailesh Kumar

Contents

There are many ways to create recommendation systems. To create an articles recommendation system, we need to focus on content rather than user interest. For example, if a user reads an article based on clustering, all recommended articles should also be based on clustering. So to recommend articles based on the content:

1.) We need to understand the content of the article 2.) Match the content with all the other articles and recommend the most suitable articles for the article that the reader is already reading. For this task, we can use the concept of cosine similarity in machine learning. Cosine similarity is a method of building recommendation systems based on content. It is used to find similarities between two different pieces of text documents. So we can use cosine similarity to build an article recommendation system. In the section below, I will take you through how to build an article recommendation system with machine learning using Python. To create an article recommendation system, I collected data about some of the articles on this website itself.

Title: India Travel Recommender System Dataset

Description

Context
Travel is a diverse and vibrant industry, and India, with its rich cultural heritage and varied landscapes, offers a myriad of experiences for travelers. The India Travel Recommender System Dataset is designed to facilitate the development of personalized travel recommendation systems. This dataset provides an extensive compilation of travel destinations across India, along with user profiles, reviews, and historical travel data. It's an invaluable resource for anyone looking to create AI-powered travel applications focused on the Indian subcontinent.

Content
The dataset is divided into four primary components:

1. Destinations: Information about various travel destinations in India, including details like type of destination (beach, mountain, historical site, etc.), popularity, and best time to visit.

2. Users: Profiles of users including their preferences and demographic information. This dataset has been enriched with gender diversity and includes details on the number of adults and children for travel.

3. Reviews: User-generated reviews and ratings for the different destinations, offering insights into visitor experiences and satisfaction.

4. User History: Records of users' past travel experiences, including destinations visited and ratings provided.

Each of these components is presented in a separate CSV file, allowing for easy integration and manipulation in data processing and machine learning workflows.

Acknowledgements
This dataset was generated for educational and research purposes and is intended to be used in hackathons, academic projects, and by AI enthusiasts aiming to enhance the travel experience through technology.

Inspiration
The dataset is perfect for exploring a variety of questions and tasks, such as: - Building a recommendation engine to suggest travel destinations based on user preferences. - Analyzing travel trends in India. - Understanding the relationship between user demographics and travel preferences. - Sentiment analysis of travel destination reviews. - Forecasting the popularity of travel destinations based on historical data.

We encourage Kaggle users to explore this dataset to uncover unique insights and develop innovative solutions in the realm of travel technology. Whether you're a data scientist, a student, or a travel tech enthusiast, this dataset offers a wealth of opportunities for exploration and creativity.

Usage

This dataset is free to use for non-commercial purposes. For commercial use, please contact the dataset provider. Remember to cite the source when using this dataset in your projects.

License

CC0: Public Domain - The dataset is in the public domain and can be used without restrictions.

The Social Recommendation Data comprises ratings alongside social or trust relationships between users from two different platforms - LibraryThing (a book review website) and Epinions (a general consumer review platform). This dataset integrates both the review data and the social connections among users, providing a unique opportunity to study how social networks influence rating behaviors and vice versa.

Basic Statistics: - LibraryThing - Number of users: 73,882 - Number of items: 337,561 - Number of ratings/feedback: 979,053 - Number of social relations: 120,536

• Epinions
  • Number of users: 116,260
  • Number of items: 41,269
  • Number of ratings/feedback: 181,394
  • Number of social relations: 181,304

Metadata: - Reviews: Textual reviews given by users. - Price Paid (Epinions): The price that users paid for the products they reviewed. - Helpfulness Votes (LibraryThing): Votes indicating the helpfulness of the reviews. - Flags (LibraryThing): Flags or tags associated with reviews or users.

Examples: - LibraryThing Reviews json { 'work': '3067', 'flags': [], 'unixtime': 1160265600, 'stars': 4.5, 'nhelpful': 0, 'time': 'Oct 8, 2006', 'comment': 'great storytelling in this novel about a couple crossed by a time travelling disorder ', 'user': 'justine' } - LibraryThing Social Network Rodo anehan Rodo sevilemar Rodo dingsi Rodo slash RelaxedReader AnnRig RelaxedReader bookbroke ...

Download Links: - LibraryThing Data - Epinions Data

Citations: - SPMC: Socially-aware personalized Markov chains for sparse sequential recommendation, Chenwei Cai, Ruining He, Julian McAuley, IJCAI, 2017. pdf - Improving latent factor models via personalized feature projection for one-class recommendation, Tong Zhao, Julian McAuley, Irwin King, Conference on Information and Knowledge Management (CIKM), 2015. pdf

Use Cases: 1. Socially-Aware Recommendation Systems: Developing recommendation models that incorporate social relationships to enhance personalization and accuracy. 2. Influence Analysis: Studying the impact of social networks on user rating behaviors and discovering influential users. 3. Trust-Based Recommender Systems: Designing recommender systems that leverage trust relationships among users for better recommendations. 4. Sequential Recommendation: Analyzing sequential behaviors in both rating and social interactions for better sequential recommendation. 5. Community Detection: Identifying communities of users with similar preferences or social connections. 6. Sentiment Analysis: Analyzing sentiments in reviews and investigating how social connections affect sentiments. 7. One-Class Recommendation: Tackling the challenges of one-class recommendation with social network information. 8. Cold-Start Problem: Addressing the cold-start problem in recommendation systems by utilizing social network data. 9. Multi-Platform Analysis: Conducting cross-platform analysis to understand the consistent and unique behaviors across different platforms. 10. Temporal Analysis: Analyzing the temporal dynamics of user behaviors and social network evolution.

This dataset is well-suited for researchers and practitioners working on socially-aware recommender systems, social network analysis, and community detection, providing a rich source of data for exploring the interplay between social interactions and user-item interactions.

We have seen many applications of Recommender systems currently like movie recommendation, book recommendation, product recommendation etc. Similarly, it is also important to have a blog recommendation system which will provide a personalized and engaging user experience by presenting relevant and interesting content to readers. By analyzing user behavior and preferences, blog recommendation algorithms can learn and adapt to each individual's interests over time, improving the accuracy and relevance of their recommendations.

By studying different recommender algorithms, we can better understand how they work and how they can be applied to different situations. For instance, some algorithms are based on collaborative filtering, which focuses on identifying similarities between users and recommending items that other users with similar interests have enjoyed. Other algorithms rely on content-based filtering, which recommends items based on the attributes and characteristics of the items themselves.

We can also explore hybrid recommender algorithms, which combine both collaborative and content-based filtering to produce recommendations that are tailored to the individual user's preferences. Through this exploration of various recommender paradigms, we can gain a deeper understanding of the complexities involved in recommendation systems and how to leverage them effectively.

Quick start

To read any dataset you can use the following code

>>> import numpy as np
>>> embed_image = np.load('embed_image.npy')
>>> embed_image.shape
(33962, 768)
>>> embed_text = np.load('embed_text.npy')
>>> embed_text.shape
(33962, 768)
>>> import pandas as pd
>>> items = pd.read_csv('items.txt')
>>> m = len(items)
>>> print(f'{m} items in dataset')
33962
>>> users = pd.read_csv('users.txt')
>>> n = len(users)
>>> print(f'{n} users in dataset')
14790
>>> train = pd.read_csv('train.txt')
>>> train
     user  item
0    13444 23557
1    13444 33739
...    ...  ...
317109 13506 29993
317110 13506 13931
>>> from scipy.sparse import csr_matrix
>>> train_matrix = csr_matrix((np.ones(len(train)), (train.user, train.item)), shape=(n,m))

Folders

This dataset contains six datasets. Each dataset is duplicated with seven combinations of different Image and Text encoders, so you should see 42 folders.

Each folder is the name of the dataset and the encoder used for the visual and textual parts. For example: bookcrossing-vit_bert.

The datasets are: - Clothing, Shoes and Jewelry (Amazon) - Home and Kitchen (Amazon) - Musical Instruments (Amazon) - Movies and TV (Amazon) - Book-Crossing - Movielens 25M

And the encoders are: - CLIP (Image and Text) (*-clip_clip). This is the main one used in the experiments. - ViT and BERT (*-vit_bert) - CLIP (only visual data) *-clip_none - ViT only *-vit_none - BERT only *-none_bert - CLIP (text only) *-clip_none - No textual or visual information *-none_none

Files per folder

For each dataset, we have the following files, considering we have M items and N users, textual embeddings with D (like 1024) dimensions, and Visual with E dimensions (like 768) - embed_image.npy A NumPy array of MxE elements. - embed_text.npy A NumPy array of MXD elements. - items.csv A CSV with the Item ID in the original dataset (like the Amazon ASIN, the Movie ID, etc.) and the item number, an integer from 0 to M-1 - users.csv A CSV with the User ID in the original dataset (like the Amazon Reviewer Id) and the item number, an integer from 0 to N-1 - train.txt, validation.txt and test.txt are CSV files with the portions of the reviews for train validation and test. It has the item the user liked or reviewed positively. Each row has a positive user item.

We consider a review "positive" if the rating is four or more (or 8 or more for Book-crossing).

The vector is zeroed out if an Item does not have an image or text.

Dataset stats

Modifications from the original source

Only a tiny fraction of the dataset was taken for the Amazon Datasets by considering reviews in a specific date range.

For the Bookcrossing dataset, only items with images were considered.

There are various other minor tweaks on how to obtain images and texts. The repo https://github.com/igui/MultimodalRecomAnalysis has the Notebook and scripts to reproduce the dataset extraction from scratch.

Dataset

This dataset was created by Vinesh Kannan

Contents

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F12038776%2F5a9c101d1a2498a37406d3a91cebb66c%2Fpkx1jz0terhb9bm50stm.jpg?generation=1713517466786485&alt=media" alt="">

Objective :

This project aims to develop a personalized course recommendation engine integrated with a Django web application, leveraging machine learning techniques. Utilizing a dataset from Udemy containing course information, the system analyzes user preferences and behaviors to provide tailored recommendations. The recommendation engine employs machine learning algorithms to predict courses that align with the user's interests based on input provided. This project demonstrates the significance of recommendation engines in enhancing user experience, increasing engagement, and driving revenue growth in the competitive digital landscape.

Dataset : * The dataset contains information on 3678 courses available on Udemy, spanning various subjects and levels of difficulty. Here's a description of the columns: * course_id: Unique identifier for each course. * course_title: Title of the course. * url: URL of the course. * is_paid: Boolean indicating whether the course is paid or not. * price: Price of the course. * num_subscribers: Number of subscribers enrolled in the course. * num_reviews: Number of reviews for the course. * num_lectures: Number of lectures in the course. * level: Difficulty level of the course (e.g., Beginner, Intermediate, Advanced). * content_duration: Duration of the course content. * published_timestamp: Timestamp indicating when the course was published. * subject: Subject category of the course. * This dataset provides comprehensive information about Udemy courses, including their popularity (measured by the number of subscribers and reviews), pricing, content duration, and level of difficulty. It covers a wide range of subjects, making it suitable for building a recommendation engine to suggest courses based on user preferences and interests.

Medicine Recommendation System using LSTM

Introduction: The Medicine Recommendation System is designed to assist medical professionals in suggesting appropriate medications for patients based on their reported symptoms and diagnosed diseases. The system utilizes Long Short-Term Memory (LSTM) neural networks, a type of recurrent neural network (RNN), to learn patterns from patient data and predict suitable medicines.

Dataset Creation: For the development and evaluation of the Medicine Recommendation System, custom datasets were created. These datasets contain records of patient cases, each comprising the patient's reported symptoms, identified disease, and the recommended medicine for treatment. The dataset is carefully labeled to ensure accuracy in training the LSTM model.

Model Architecture: The core of the system is an LSTM neural network. LSTM networks are well-suited for sequence-based data, making them a suitable choice for analyzing patient symptoms and diseases. The LSTM architecture allows the model to capture long-range dependencies in the patient's reported symptoms and predict suitable medicines based on these patterns.

Training Process: The LSTM model was trained on the created datasets using a supervised learning approach. The patient symptoms, identified diseases, and recommended medicines serve as input-output pairs for the training process. The model optimizes its internal parameters to minimize the difference between its predictions and the actual recommended medicines. During training, the accuracy of the model was observed to be approximately 88%.

Working of the System:

Input: The user provides the system with the patient's reported symptoms, identified disease, and any relevant information. Processing: The LSTM model processes the input, leveraging its learned patterns to predict the most suitable medicine for the patient's condition. Output: The system presents the user with the recommended medicine for the patient based on the input data. Accuracy and Performance: The trained LSTM model demonstrates an accuracy of around 88% during testing and validation. This accuracy indicates the model's ability to correctly predict the appropriate medicine based on the provided input symptoms and disease.

Future Enhancements:

Data Augmentation: Enhance the dataset by incorporating a wider range of patient cases to improve the model's generalization capability. Fine-Tuning: Experiment with hyperparameter tuning and model architecture adjustments to potentially improve prediction accuracy. Integration with EHR: Integrate the system with Electronic Health Records (EHR) systems to provide real-time recommendations based on patient history. Conclusion: The Medicine Recommendation System, built on an LSTM architecture, shows promise in accurately predicting suitable medicines for patients based on their reported symptoms and diagnosed diseases. With an accuracy of around 88%, the system demonstrates the potential to assist medical professionals in making informed medication decisions, contributing to enhanced patient care and treatment. Further improvements and integrations could unlock even more capabilities and benefits for healthcare practitioners.

The datasets were collected in late 2017 from goodreads.com by "hashank Kapadia", where the data from users’ public shelves was scraped, i.e., everyone can see it on the web without login. User IDs and review IDs are anonymized.

There are two groups of datasets:

• meta-data of the books
• user-book interactions

These datasets can be merged together by matching book/user/review ids to apply different methods and algorithms to build a recommendation system.

A book recommendation system is a type of recommendation system where we have to recommend similar books to the reader based on his interest. The books recommendation system is used by online websites which provide ebooks like google play books, open library, good Read’s, etc.

During the last few decades, with the rise of Youtube, Amazon, Netflix and many other such web services, recommender systems have taken more and more place in our lives. From e-commerce (suggest to buyers articles that could interest them) to online advertisement (suggest to users the right contents, matching their preferences), recommender systems are today unavoidable in our daily online journeys. In a very general way, recommender systems are algorithms aimed at suggesting relevant items to users (items being movies to watch, text to read, products to buy or anything else depending on industries). Recommender systems are really critical in some industries as they can generate a huge amount of income when they are efficient or also be a way to stand out significantly from competitors. As a proof of the importance of recommender systems, we can mention that, a few years ago, Netflix organised a challenges (the “Netflix prize”) where the goal was to produce a recommender system that performs better than its own algorithm with a prize of 1 million dollars to win. By applying this simple dataset and related tasks and notebooks , we will evolutionary go through different paradigms of recommender algorithms . For each of them, we will present how they work, describe their theoretical basis and discuss their strengths and weaknesses.

There are a lot of applications where websites collect data from their users and use that data to predict the likes and dislikes of their users. This allows them to recommend the content that they like. Recommender systems are a way of suggesting or similar items and ideas to a user’s specific way of thinking.

This dataset is perfect if you want to try building a movie recommendation system.

Dataset

This dataset was created by Vipin_yadav_1809

Released under CC0: Public Domain

Contents

Dataset

This dataset was created by Sowjanya Lahari Nagidi

Released under CC0: Public Domain

Contents

Dataset

This dataset was created by Tanuj dhiman

Contents

Context

These files contain metadata for over 20,000 movies listed in the Full TMDB Dataset. The dataset consists of movies released on or before August 2022 as well as some of the upcoming movies till Dec 2028. Data points include title, release dates, languages, genre, popularity, TMDB vote counts, and vote averages.

Acknowledgements

The Movie Details have been collected from the TMDB Open API. This product uses the TMDb API but is not endorsed or certified by TMDb. Their API also provides access to data on many additional movies, actors and actresses, crew members, and TV shows. You can try it for yourself here.

Inspiration

This dataset is assembled as part of my Project for Recommender Systems. I wanted to perform an extensive EDA on Movie Data to build various types of Recommender Systems.

Dataset

This dataset was created by Raihan Sikdar

Contents

# appid: Unique identifier
# name: Name of the game
# relase_date: release date
# english : Is there English language support
# developer : Developer Company
# publisher : Publisher Company
# platforms : supported platforms; windows, mac, linux
# required_age : Is 18 years old required
# categories : game categories
# genres : type of game
# steamspy_tags : Game Genre Tag
# positive_ratings : number of up comments
# negative_ratings : Number of down comments
# average_play_time : average time played
# median_playtime : median time played
# owners: estimated user range
# detailed_description : detailed description about the game
# about_the_game: general description about the game
# short_description: short description about the game