# Databases for bibliography and state of the art
## Databases
Bibliography and state of the art : in any PhD thesis discovering and working with bibliographical sources is necessary.
This is made possible by databases that let users perform systematic searches based on well-thought criteria.
At Université Paris-Saclay 3 main databases are used for this:
- Scopus
- Web of Science
- OpenAlex
Scopus and the Web of Science are subscription-based, curated databases for bibliographic and bibliometric data. This means that: 
- Their data is not open, but it is richer than otherwise, and that any Université Paris-Saclay student or staff can use them for free thanks to the subscription.
- They gather sources mainly from journals and conferences with international audience (English language), that meet quality criteria defined by experts.
- They are made to search for metadata, including titles, abstracts, publication years, author names, citations.
OpenAlex is an uncurated bibliographic an bibliometric database.
- Their data is fully open, licensed as CC0 (public domain), for anyone to use.
- They gather as many academic sources from all countries as they can, no matter the perceived quality, so that users make the choice rather than themselves.
- Creating an account lets users get an API key for automated queries and easier AI-based use.

## Using OpenAlex for bibliographic work
### Search modes in OpenAlex
Basic search from UI is simply a keyword search on titles and abstracts.

Advanced search lets user choose their criteria and operators between them (AND, OR, either/or possibilities).

There is a complete query language named OQL, pretty similar to simple English, but it needs to be learned or written by AI.

All the documentation and help in OpenAlex is easy to access, API doc included. There is an "API" button in all result pages to be able to see the full json data behind any page, and to help user build their API calls if needed.

### Tips for bibliographic work
Reading all the potentially relevant literature for any PhD thesis subject is almost impossible because there is too much. Bibliography means people have two steps to perform:
1. Find a way to gather all the relevant literature in one place, e. g. one databases query that they can execute periodically for updates.
2. Find a way to then filter all the papers they found with criteria they define:
	1. Relevance, sometimes by narrowing first search terms to narrower terms that can change with time
	2. Publication types, by defining if some types of publications (e. g. journal articles, canference papers, posters, book chapters...) are more or less relevant to what they are trying to do at a given moment
	3. Publication years, by defining if some older works are no longer relevant for any reason (already read, technology has evolved, subject did no exist at the time...) or if they want to focus on older works, for example while trying to understand the fundamentals of a subject.
	4. *Citations*, by using a set of important works as a base for building a citation graph and using what the set cites or what cites the set as a primary extension of it. Citation numbers can also be used as a proxy for quality, meaning sometimes works with fewer citations are less important to read (carefull with this criterion, it is not always true)
Once both steps are defined, methodology used for bibliography can be written down and it serves as a quality indicator for the state of the art.