I am a highschool student working on a science fair project in which me and a friend plan to use a neural network for a classification problem. In this case the thing classified will be text and the labels will be the publication/creation date. Since our project requires a significant amount of data we were planning to use the Gutenberg project. However we came upon the issue that the creation date was not in each file but rather on a website associated with the name (so we have to go to a website which ends with the text's file name). We are wondering what the best approach would be to get our data - should we somehow get the date using the website (even though they say they will ban ip's of scrapers), find a new dataset, or build some sort of tool which will search the internet for that text's creation date. Thanks and if you have any suggestions please feel free to comment. If our question should be further expanded upon please clarify. Thanks again.
1 Answer
I would hazard a guess that most books uploaded on Gutenberg also have a Wikipedia page. Especially the well-known books. There are a few Wikipedia related Python packages (i.e. Wikipedia) that can make it very easy to search for the title of the book, navigate to the book's page, and extract the publication date. An optimal example might be Alice's Adventures in Wonderland. It'd be trivial to extract the date from that page, which appears neatly in the info box, and convert it to a machine readable format in Python.
Or, you could attempt to scrape the info straight from Gutenberg. Perhaps consider Selenium, although make sure you limit your bot's rate so you don't adversely effect the site.
Lastly, trying to predict the publication date might be similar to predicting the author; both may be derived from writing style. There's a lot of work on this topic, a Google search or an arxiv search will help. Here's one of the first articles I opened.