The Web of Science (WOS) is a database produced by Thomson Reuters and is accessible with a subscription. It is the reference resource for the evaluation reports of research units, reports from the Observatory of Science and Technology (OST), Shanghai rankings, Times Higher Education rankings, and more. It is not without its critics: essentially Anglo-Saxon coverage, near-absence of the human and social sciences, economy and finance, computer science, mathematics and certain engineering domains, and little processing of affiliations and cited references.

The list of journals indexed by the WOS is organized by domain:

Social Sciences Citation Index
Arts and Humanities Citation Index
Science Citation Index EXpanded

WOS journal selection process: http://wokinfo.com/essays/journal-selection-process/

Scopus is a competitor to WOS, produced by the Elsevier publisher and accessible with a subscription. The database offers the same functionalities (for example, analysis of the results by affiliation, by country, ranking by number of citations, searching in the cited references, etc.) and the same types of indicators as Thomson Reuters (number of citations per article, h-index, average number of citations per author...).

As in the WOS, the database is clearly identified: the list of indexed journals is specified (http://info.sciencedirect.com/scopus/france/) and this is important for the analysis. As in the WOS, the data (author addresses, author names, cited references...) are transcribed as they appear in the source journals. Although some "data cleansing" is done and the producers of these databases encourage authors to report errors, the data are not necessarily homogeneous and complete, which means that the analysis results are relative.

Although Scopus includes more journal titles than the WOS, particularly in the human and social sciences, the citation period is currently shorter, because the database only incorporates references cited after 1996 (compared to 1900 for the WOS or 1975 for current subscriptions). These are today the two main differences between these two commercial databases.

The Journal Citation Report compiles several journal indicators organized around the impact factor: immediacy index, cited half-life, quartiles…
The report is produced by Thomson Reuters every year and is only accessible with a subscription.

Essential Science Indicators (ESI) is another commercial product from Thomson Reuters, which uses the citation data in the WOS and extracts the most cited countries, institutions, authors and journals, averaged over 10 years, in the main disciplinary domains defined by the database. All countries, institutes... do not appear in all the domains in the rankings. For example, for an institution to be integrated in the institute rankings, the cumulative number of citations should be above the threshold of 1% of the number of total citations in the given domain.

This tool also highlights the "most often cited" publications (highly cited, top 1%) by domain over the last 10 years and those that have been "the most prominent" (hot papers, top 0.1%) by domain, i.e. those that have received the largest number of citations, the fastest, over the last 2 years.

Some ESI data are compiled and published with open access on the Science Watch website : http://archive.sciencewatch.com/dr/rfm/

SIGAPS (Système d’Interrogation, de Gestion et d’Analyse des Publications Scientifiques) is a French database for the collection and analysis of publications, implemented for all French University Hospitals (CHU), based on data from the Medline database. Access is restricted.

Google Scholar is Google’s  "scientific" navigator. It displays citations from scientific publications across all websites indexed by the browser. It also proposes a personal reference tool: My citations.

Google Scholar has the advantage of being free and covers publications in all domains and in all forms (including conference communications, reports, patents, theses and books). For the calculation of its indicators, a certain amount of selection is done: http://scholar.google.fr/intl/fr/scholar/metrics.html#coverage.

Nevertheless, and this is the big difference between Google Scholar and the WOS or Scopus, the exact limits of the sources taken into account in the calculation of citations is not known. There is no journal selection, and no guarantee that the list is stable. There is no processing of duplicates, triplicates... etc. It does not allow searches on affiliations, like the WOS or Scopus.

Faculty of 1000 is a commercial database specialized in the biomedical domain, which proposes notably a selection of articles that are commented and evaluated by a network of scientists and a ranking of these articles by domain (F1000Prime).

CiteseerX is an open access archive specialized in computer science and applied mathematics, which indexes citations and offers rankings. It is constructed by harvesting publications from free web sites and therefore does not include publisher websites that are only accessible with a subscription.

Citebase (beta) calculates the number of downloads and the time to citation based on open access archives (ArXiv, ECS EPrints, Soton EPrints). It is specialized in the physics and astronomy domains.

Publish Or Perish (PoP) from Harzing is free software that uses data from Google Scholar. It thus covers all domains and all types of publications, but also has the same limitations as Google Scholar. PoP publishes many bibliometric indicators: http://www.harzing.com/pop.htm#about

SCImago Journal & Country Rank is an open portal for bibliometric analyses produced using data from Scopus. It hosts the data for the SJR (SCImago Journal Rank) indicator.