Document-driven investigative tools
This is a collection of tools for ingesting, normalizing, indexing and tagging documents in the context of a journalistic investigation.
- As a journalist, I want to store a list of documents that mention a person/org/topic so that I can sift through the documents.
- As a journalist, I want to intersect sets of documents that mention people/orgs/topics so that I can drill down on the relationships between them.
- As a journalist, I want to combine different types of facets which represent document and entity metadata.
- As a data importer, I want to routinely crawl and import documents from a data source.
- As a data importer, I want to associate metadata with documents and entities to allow advanced facets.
- An entity (such as a person, organisation, or topic) is always a search query; each entity can have multiple actual queries associated with it by means of aliases (tags?).
- Documents can be anything, and there is no guarantee that
ditwill be able to display it - just index it. Document display is handled by DocumentCloud etc.
- Documents matching an entity after that entity has been created yield notifications if a user is subscribed.
- nltk, patterns
- OpenCalais, LingPipe, AlchemyAPI
dit uses textract, which has external (i.e. non-Python) dependencies. See the install guide.
apt-get install python-dev libxml2-dev libxslt1-dev antiword poppler-utils pstotext tesseract-ocr flac ffmpeg lame libmad0 libsox-fmt-mp3 sox
aleph is licensed under a standard MIT license (included as