[from Chris at OpenCorporates]
Thanks for all the useful comments. Wondered if anyone tried the Advanced Search, and particularly the experimental 'search by relevance', which gives Apple Inc as the top score. It's experimental in part because what two users my not have the same view of relevant as each other (one reason why Google wants you to be logged in when searching, so it can return searches on what it thinks is relevant to you).
Re the number of results, with over 95 million companies from hundreds of official sources, there are a lot of companies with the same or similar names -- that's part of the power, and why OpenCorporates is so widely used by journalists, anti-corruption investigators, lawyers, law-enforcement etc. The more we loosen the search, however, the more results you will get -- so it's a balancing act, and the advanced search and the API is part of us trying to make it work for users, and we'd love feedback on both -- just email us at community @ opencorporates dot com
For large listed companies such as Apple or Intel, it depends on what question you are trying to answer. If it's just an overview of the corporation, then Wikipedia or something like Yahoo Finance is the best route. The former gives a narrative overview, using the collected expertise of hundreds of contributors; the latter includes highly proprietary data (which was also for the most part collected by offshore humans) to build an overall picture.
However, for official, provenanced data under an open licence about legal entities, OpenCorporates is by far the best option -- the Facebook example is a good one. Contrary to many people's expectations, Facebook is not a California Corporation. However, because they operate in California they have to register as a branch (aka Foreign Corporation), and if you land on that page (https://opencorporates.com/companies/us_ca/C2711108) you'll see that we actually link to the home corporation, in Delaware (https://opencorporates.com/companies/us_de/3835815), and there we list subsidiaries and other branches. Why don't we have Instagram? You can check from the source of the subsidiaries we list (e.g. https://opencorporates.com/statements/37307675 which links to the SEC Exhibit 21 filing at http://www.sec.gov/Archives/edgar/data/1326801/0001326801150... ), and you can see there that they don't list Instagram (possibly because either there's no separate company for it, or because it doesn't count as a material subsidiary).
So the question often becomes (and news stories are problematic for multiple reasons), where can you find the linkages that can be parsed into structured, provenanced data in a reliable way. We're focusing on two areas: looking for sources of public data that can be combined together to give insight to all, and building up a community of users to help us do that: http://impact.opencorporates.com/contribute/
Please do consider joining us in this important mission, or just by pinging us with suggestions of how we can do better.
[Chris from OpenCorporates here]: Any code you write is your code... and we'd be happy for you to share it by whatever means you want, e.g. by pushing to a public github/gitlab repo.
We did originally plan to do this by default, but in an earlier version of this work some contributors said that they didn't mind OpenCorporates having the benefit, but didn't want proprietary business information companies to have it.
We've got a ticket to add to the manifest and option public_repo field, so that the location of the public repo is easily discoverable, and hope to implement in the next week or so.
The share-alike refers to the underlying data, but again, if you want to make it available to the world under a CC0 licence we have no problem with that.
[Chris from OpenCorporates] This mission is specifically about bank and financial licences, as opposed to physical bank branch locations, which is slightly different (though related). We think this dataset is an essential foundation for such datasets, and others. It also links the banks to company records, and from that to potentially other data, from political contributions/lobbying to corporate networks (see https://opencorporates.com/viz/financial/index.html -- some of this data came from the National Information Center, which is a joint project of the US banking regulators).
However, if you think there's a company-related government dataset that should be scraped, please do go ahead and use turbot.opencorporates.com for that -- it's already been used for everything from civil aircraft registrations to mining licences to NHS Providers.
Hi, Chris from OpenCorporates.
We're also getting the native classifications, i.e. those given by each regulators, and in a sense these are mappings. When researching this, the are many, many different classifications (particularly in the consumer finance and shadow banking area), and one of the side benefits this will surface those classifications, laying one of the foundations for a more formal ontology.
The wider problem with the DUNS number, especially as far as its use within government, is that it is a proprietary identifier, which makes it unsuitable for use in a public domain context, especially as far as combining with other data (e.g. combining data from different federal agencies) or in an open data context. Having said that the UK government is now using it in procurement just as the US government is now starting to understand the underlying problems and having second thoughts about its use. See http://blog.opencorporates.com/2012/07/24/are-duns-numbers-t... for more details