Application usage
Application by default runs on http://127.0.0.1:5000 in case of running on external server all links in this instruction must be aligned to server address.
Downloading articles may take a long time, even about 30s per article, that's why there are two ways of initiating download.
Downloading using scripts is recommended when one intends to download large number of files. To do it you shall type following command:
$ run_articles_download.py --file_type=FILE_TYPE --doi_list=INPUT_FILE --output_articles=SERVER_WORKSPACE_DIRECTORY/.server_files/articles
Where INPUT_FILE is a file used for download of given FILE_TYPE and SERVER_WORKSPACE_DIRECTORY is a directory where flask run was called.
Information about input files can be found in here, the INPUT_FILE values can be checked by running :
$ run_articles_download.py --help
This will prepare articles text, so that next you can run finder on them using web-ui without additional overhead for downloading.
To download and run finder using web-ui one has to visit http://127.0.0.1:5000/upload and select articles csv file and either finder as a txt file or as a text and press Process Articles
Warning! This may take a long time if articles were not already downloaded, and a web browser will show page loading only. It is recommended to use this way to run finder on articles and when it is expected that only few new articles will be downloaded (e.g. after using script)
Users can register under http://127.0.0.1:5000/register and login under http://127.0.0.1:5000/login. Being logged in allows user to add comments and accept/reject articles (to further reading)
User can view articles in a convenient form.
One can view articles in a list form under http://127.0.0.1:5000
The article can have one of following statuses:
- Articles in which finder found a match and the article text was properly downloaded
- Articles in which finder found a match (e.g. in abstract) and the article text could not be downloaded
- Articles in which finder did not find a match and application could not get their full text because it was not available or some error occurred
- Articles in which finder did not find a match and download of full text from this publisher is not supported
- Articles in which finder did not find a match and that were properly read (full text available)
- Articles ignored e.g. some "Proceedings" articles which or articles marked as not related by user
User can also expand information about article to see its doi and a list of statuses (ACCEPTED/REJECTED/TO BE CHECKED) for each user.
A view of each article has two parts:
First part contains title, read status, authors and other bibliographic information on the left side. Users can also mark this article as ignored - e.g. because it does not relate to given field.
On the right side there is a panel for comments and buttons to ACCEPT/REJECT article (availiable for logged users)
Second part contains full text of articles, articles are split into sections and initially only sections which contains any finder match are expanded. Additionally only sentences and paragraph with matches are expanded. It is possible to see the rest of text using buttons inside text.
To generate results one shall click Generate Results http://127.0.0.1:5000/results in navigation panel. This will cause application to prepare output in csv format containing Titles, DOIs, Authors and Statuses from each user.