You cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.
\n\n Privately sharing code or data outside of teams is not permitted.\n It's okay to share code if made available to all participants on the forums.\n
\nTeam mergers are not allowed in this competition.
\n\nThe maximum team size is 2.
\nYou may submit a maximum of 3 entries per day.
\nYou may select up to 1 final submissions for judging.
\nStart Date: 3/1/2018 22:00 PM UTC
\nEnd Date: 3/11/2018 23:59 PM UTC
\n\t", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "data-description", "content": "We retrieved these documents from the internet, where both companies are mentioned.\nThe dataset is split into 2 parts, one folder for each.\n\n## File descriptions\n\n - **files.zip** - Contains two folders, one for each company's scraped data\n - **sample_submission.csv** - Use this format to submit your results\n - **submission_mapper.csv** - Use this file to map each submission `id` to each file and its corresponding company\n\n## Data fields in the submission mapper file\n - **id** - The `id` of each file/company combo\n - **filename** - The `name` of each file\n - **company** - Company associated to the file (there can be 2 lines for the same file, one for each company)\n\n## Data fields in the submission file\n - **id** - The `id` corrsponding to the mapper's `id`\n - **rank** - The ranking of each file for its company, comprised in `[-10,10]`", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Description", "content": "## Motivation ##\nPerhaps the most critical, strategic, and an enduring asset that a corporation possesses is its reputation. Although corporate reputation is undoubtedly a significant and relevant corporate asset, formidable measurement challenges exist nowadays. We propose the creation of a \"reputation index\" based on public articles and all media resources, which are accessed on the internet and are relevant to the particular company.\n\nOur approach is based on a Sentiment analysis strategy to evaluate all the events related to the company and its employees, client's opinion and calculate the general index, which can define the level of the reputation.\nSeveral studies have focused on the same topic of the sentiment changes overtime in online media. Gilbert & Karahalios (2010) have estimated information about future stock market prices based on the analysis of the emotions expressed in blogs. Based on over 20 million posts made on the site LiveJournal, they found that increases in expressions of anxiety predict downward pressure on the S&P 500 index. O´Connor et al. (2010) have found that a relatively simple sentiment detector based on Twitter data replicates consumer confidence and presidential job approval polls.\n\n## Task ##\nWe propose to combine Sentiment analysis and Entities recognition methods to find the 10 most positive and 10 most negative files about each of the 2 proposed companies: Keppel (Keppel Corporation Limited) and Prudential (Prudential Assurance Company).\n\n## Hints ##\nNatural Language Processing methods as: Sentiment analysis, Entities recognition and Text summarization should be used in this competition.\n\n## Evaluation ##\nIn order to explore the feasibility of automating reputation analysis processes, we have implemented a prototype system that calculates the reputation index in newspaper articles.\nThe prototype has been used to find the files with the best/worst opinion about the companies.\n\n## Team ##\nWe will only consider teams of 1 or 2.\n\n## Prize ##\n**To be eligible for prizes, you must be an EPFL student, or have graduated from EPFL in the last 6 months as this is an internal competition.**\n\n - The winning team will be awarded 2 return tickets from Geneva to Singapore. But also, if you so choose to apply for an internship/master's project with us at **Dathena**, we will consider that you passed the technical interview, hence you will only interview with the HR department.\n - The 2nd best team will be awarded 2 Bose Headphones\n - The 3rd best team will be awarded a Nespresso machine\n\n## Questions ##\nIf you have any questions, please send an email to `romain.choukroun@dathena.io` or `guillaume.mollard@dathena.io`", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Evaluation", "content": "\nOne of the most important tasks of a Data Scientist is to communicate their results. This is why we will mostly evaluate your results by taking a look at your pipeline. To do so you will need to submit here your ranking file, but also to send us by mail a zip folder containing your code and a notebook/pdf explaining your pipeline. \n\n## Scoring\nThe ranking is in `[-10,10]`, `0` meaning **not in the top 10 best/worst**, with `-10` being the worst, `10` being the best, and so on. Which means that you should have `40` non-zero rankings, `20` for each company. The score is computed as a MAE with parameters your predictions and ours, hence the smaller the score, the closer you get to our solution.\n\nThese scores should serve as indications that you are in the right track. **Keep in mind that we will mostly base our evaluation on your code and pipeline.**\n\n## Submission Format\nThe score submission should be a csv file with two columns, `id` and `rank` separated with a `,`, and comprising the header on the first line. Use the sample submission file by only replacing the dummy rankings and you should be fine. Each `id` is mapped to a file/company combo, find the mappings in the `submission_mapper` file\n\nThe pipeline submission should be a zip file containing your code in any chosen language, and a notebook or pdf explaining your pipeline, of course it should also contain your team name, team members and respective email addresses.\n\nSend your zip submission by mail to `romain.choukroun@dathena.io` and `guillaume.mollard@dathena.io` within 24 hours of the deadline. **Your participation will only be taken into account if you both submit the csv file on Kaggle and also send us the zip.**", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false } ] }