Internet addiction refers to excessive internet use that interferes with daily life. Due to its negative impact on college students’ study and life, discovering students’ internet addiction tendencies and making correct guidance for them timely is necessary. In this system, we provide an approach to estimating a person’s internet addiction levels using their behavior data on the system. In detail, we consider a person’s addiction to the internet as a hidden variable that affects students’ daily time online together with other behavior. By predicting a person’s daily time online, we will find a person’s internet addiction levels. Finally, extensive experiments are conducted on a real- world dataset. The experimental results show the effectiveness of our method, and it is also consistent with some psychological findings. “I-detox” aims to provide real-time analysis of the exact number of websites visited by the user, giving them the freedom to foresee the probable results given the lack of schedule one follows while browsing the internet. Therefore, this gives a standardized analysis of the web history of the user also inclining the number of times a user visits a particular website. Our solution considers the suitability of terming the user as an addict or not based on the analysis provided upfront.
Addictive digital experiences are becoming extremely common in the tech industry and the adjective “addictive” is now considered one of the highest compliments paid to an artifact. Existing research has explored what is considered addictive, what are the approaches and strategies performed by tech companies that eventually lead to addiction, and how cultural and societal pressures have contributed to this phenomenon. Addiction to technology has grown into a fully recognized disorder . “I-detox” can be used from the young to the old because, in the age of digitalization, one never knows the extent of addiction one is suffering. Our solution gives the user a clear idea about the number of websites being seen, the type of content being surfed upon, the category it belongs to, and the URL.
Our motivation is to aim and target the new age internet addiction. This addiction can spread from the young to old and can be equally dangerous no matter what age group it subsides too. Our application aims to help individuals who want to analyze their browsing pattern, and decide the chances of addiction..
III. LIMITATION EXISTING SYSTEM OR RESEARCH GAP
On analyzing the current systems existing in the market, we find that the system’s do track their history into a file/text document but don’t go further to categorize it. This may serve as a problem since not every website can be identified by its URL. “I - detox” not only provides the history information into an email but also categorizes the data into different types of websites adding onto the display of bag of words searched on a particular website. Thus, this makes the system user-friendly providing clarity which website provides what purpose and service and further analyzing how much time was spent on a website. Browsing history/ or what does the user access internet for can prove a major aspect to analyze the human-being as a whole and what he/she is thinking. On analyzing one can find if he/she is facing some mental problems. Currently, no system or software goes out of the way to analyze a user’s browsing history and analyze their browsing pattern or so. That’s what “I-detox” brings to the table.
IV. LITERATURE SURVEY
A study on rough set-aided feature selection for automatic web-page classification by Wakaki et.al gives a basic idea about how the web-pages are classified based on categories. 
Keras: The Python Deep Learning library by Son Eris¸im is a well built python library which makes our thoughts clear on the concepts of deep learning and provides an API for implementing it. 
A study on Internet addiction and its relation to psy- chopathology and self-esteem among college students by Kumar M et.al gives basic ideas about Internet use and its relation to psychopathology and self-esteem among college students.
Large Scale Text Classification using Semi-supervised Learning Multinomial Naive Bayes Jiang Su,Jelber Shirabad,Stan Matwin A research paper which clearly specifies how multi-novial Naive Bayes can be used for effective classification with outstanding results. 
Internet Addiction Facts and Statistics By The Recovery Village. By Megan Hul. Medically Reviewed By Dr. Andrew Proulx, MD helps us to understand what addiction is and how it is caused. 
Text classification techniques: A literature review By M. Thangaraj, M. Sivakami. A paper which explains the back- ground of various classification techniques their uses way of functioning and their implementation .
Internet addiction and problematic Internet use: A system- atic review of clinical research by Daria J Kuss and Olatz Lopez-Fernandez gives a comprehensive overview of clinical studies on the clinical picture of Internet-use related addictions from a holistic perspective. A literature search was conducted using the database Web of Science. 
V. METHODOLOGY APPLIED
To provide the user with his/her browsing history through email. The data would be processed. Basic analysis will be performed on the data. The system would be trained for analysis, the training data would be regressed and passed to the user. Model would be assessed using fresh test data. Training set: This set of data must be used to train the model for it to accurately provide the search history. Testing set: The model will be tested using some test data after it has been trained. After that, the training data will be passed into a regressor, which is used to predict continuous values.  Finally, our model will be assessed using fresh test data.
A. Specific Requirements
a. Language: Python
b. IDE: Spyder
c. ML Model Platform: Colab Hardware Requirements:
d. 2Gb Ram
e. Intel Pentium Gold Processor and above / Ryzen Athlon and above
f. 512 Mb Physical Storage and above
a. Multinomial NB: An technique to Bayesian learning that is well-liked in Natural Language Processing (NLP) is the Multinomial Naive Bayes algorithm. Using the Bayes princi- ple, the programme makes an educated prediction about the tag of a text, such as an email or news article. It determines the likelihood of each tag for a particular sample and outputs the tag with the highest likelihood. Each feature being classified by the Naive Bayes classifier is distinct from every other feature, which unites the various algorithms that make up the classifier. The inclusion or removal of one characteristic does not depend on the presence or absence of another feature. The Naive Bayes method is an effective technique for examining text input and resolving issues involving several classes. It is vital to first understand the Bayes theorem concept because the Naive Bayes theorem is built on it. The Bayes theorem, which was created by Thomas Bayes, calculates the probability of an event happening based on information about its circumstances. We determine the chance of class A when predictor B is available. Its foundation is the following equation: P(A—B) = P(A) * P(B—A)/P(B). Because all that is required to accomplish it is probability calculation, it is straightforward. Both continuous and discrete data can be used in this strategy. Real-time applications can be predicted using this simple method. Large datasets may be handled with ease and it is highly scalable. The prediction accuracy of this method is less accurate than other probability algorithms. Regression is not applicable. It is impossible to estimate numerical values using the Naive Bayes technique; it can only be used to classify textual input.
b. Beautiful Soup: Beautiful Soup is a Python library for pulling data out of HTML and XML files. It works with your favorite parser to provide idiomatic ways of navigating, searching, and modifying the parse tree. It commonly saves programmers hours or days of work.
3. Website Classification
First, the local system is used to access the web browser history. The URL, the number of visits, and the time stamp (the most recent visit) are all included in the browser history. The model categorises the urls using this web browser history. The URLs are categorised using lovely soup, which uses the url’s html description to do so. The user report is produced for a period of one month based on this categorisation. The concerned party might examine user behaviour after receiving the report.
4. Count Vectorizer
Count Vectorizer is a great tool provided by the scikit-learn library in Python. It is used to transform a given text into a vector based on the frequency (count) of each word that occurs in the entire text. This is helpful when we have multiple such texts, and we wish to convert each word in each text into vectors (for using in further text analysis)
Web scraping is done in this section using a technique called beautiful soup. We ask the user to enter the website url first. Once the url has been acquired, it is split on the third ’/’ and the remaining portion is split across the ’/’ before being joined to provide the parsed url that has been optimised. The header for the URL is taken based on the browser once the url has been acquired. The header is obtained from the website-input and added to ”r,” after which the html page is obtained from the content of ”r.” Metadata is present on the HTML page. The categorisation is based on this meta data.The ’description’ element, which holds the text for the url classification, is now acquired from the meta data. The description’s data is then transformed into an array and saved as an array. The model is then given this content to determine the website’s categorization. The anchor element that contains the url link makes up the link. This link is included in the material and kept there. The contents of the url itself are broken and kept as an array and utilised for prediction if there is a website with security measures that make it difficult to access the meta data. As a result, the website is classified in this way. Additionally, a ”cannot access website” warning is displayed if a website is inaccessible.
In this project, we propose an artificial intelligence-based system that can help the recipient analyze for himself his use of the internet. This system can work with people of all age groups with parents using it to keep a check on their ward’s internet life to the people who find themselves lost in this huge digital world. This system would help the user have a personalized analysis and a fact check about his presence on the internet, the activities he performs, the content he streams to the games he plays because of the availability of timestamps recording details of every thin activity. Thus, the question in person with the help of our system can reconcile his use of the internet thus helping him curb and storm out his way through these addiction demons.
 Wakaki, T., Itakura, H., Tamura, M., Motoda, H., Washio, T. (2006). A study on rough set-aided feature selection for automatic web-page classification. Web Intelligence and Agent Systems: An International Journal, 4, 431–441
 Keras: The Python Deep Learning library, https://keras.io/, Son Eris¸im: May?s 2018.
 Large Scale Text Classification using Semi-supervised Multinomial Naive Bayes Jiang Su, Jelber Shirabad, Stan Matwin https://icml.cc/2011/papers/93icmlpaper.pdf Text classification techniques: A literature review. M. Thangaraj, M. Sivakami https : //www.proquest.com
 Internet Addiction Facts and Statistics by The Recovery Village —Editor Megan Hull
 Internet Addiction: Definition, Assessment, Epidemiology and??Clinical Management by Donald Black.