Linked Literature, Linked TV – Everything Looks like a Graph

cloud

Ben Fry in ‘Visualizing Data‘:

Graphs can be a powerful way to represent relationships between data, but they are also a very abstract concept, which means that they run the danger of meaning something only to the creator of the graph. Often, simply showing the structure of the data says very little about what it actually means, even though it’s a perfectly accurate means of representing the data. Everything looks like a graph, but almost nothing should ever be drawn as one.

There is a tendency when using graphs to become smitten with one’s own data. Even though a graph of a few hundred nodes quickly becomes unreadable, it is often satisfying for the creator because the resulting figure is elegant and complex and may be subjectively beautiful, and the notion that the creator’s data is “complex” fits just fine with the creator’s own interpretation of it. Graphs have a tendency of making a data set look sophisticated and important, without having solved the problem of enlightening the viewer.

markets

Ben Fry is entirely correct.

I suggest two excuses for this indulgence: if the visuals are meaningful only to the creator of the graph, then let’s make everyone a graph curator. And if the things the data attempts to describe — for example, 14 million books and the world they in turn describe — are complex and beautiful and under-appreciated in their complexity and interconnectedness, … then perhaps it is ok to indulge ourselves. When do graphs become maps?

I report here on some experiments that stem from two collaborations around Linked Data. All the visuals in the post are views of bibliographic data, based on similarity measures derrived from book / subject keyword associations, with visualization and a little additional analysis using Gephi. Click-through to Flickr to see larger versions of any image. You can’t always see the inter-node links, but the presentation is based on graph layout tools.

Firstly, in my ongoing work in the NoTube project, we have been working with TV-related data, ranging from ‘social Web’ activity streams, user profiles, TV archive catalogues and classification systems like Lonclass. Secondly, over the summer I have been working with the Library Innovation Lab at Harvard, looking at ways of opening up bibliographic catalogues to the Web as Linked Data, and at ways of cross-linking Web materials (e.g. video materials) to a Webbified notion of ‘bookshelf‘.

In NoTube we have been making use of the Apache Mahout toolkit, which provided us with software for collaborative filtering recommendations, clustering and automatic classification. We’ve barely scratched the surface of what it can do, but here show some initial results applying Mahout to a 100,000 record subset of Harvard’s 14 million entry catalogue. Mahout is built to scale, and the experiments here use datasets that are tiny from Mahout’s perspective.

gothic_idol

In NoTube, we used Mahout to compute similarity measures between each pair of items in a catalogue of BBC TV programmes for which we had privileged access to subjective viewer ratings. This was a sparse matrix of around 20,000 viewers, 12,500 broadcast items, with around 1.2 million ratings linking viewer to item. From these, after a few rather-too-casual tests using Mahout’s evaluation measure system, we picked its most promising similarity measure for our data (LogLikelihoodSimilarity or Tanimoto), and then for the most similar items, simply dumped out a huge data file that contained pairs of item numbers, plus a weight.

There are many many smarter things we could’ve tried, but in the spirit of ‘minimal viable product‘, we didn’t try them yet. These include making use of additional metadata published by the BBC in RDF, so we can help out Mahout by letting it know that when Alice loves item_62 and Bob loves item_82127, we also via RDF also knew that they are both in the same TV series and Brand. Why use fancy machine learning to rediscover things we already know, and that have been shared in the Web as data? We could make smarter use of metadata here. Secondly we could have used data-derrived or publisher-supplied metadata to explore whether different Mahout techniques work better for different segments of the content (factual vs fiction) or even, as we have also some demographic data, different groups of users.

markets

Anyway, Mahout gave us item-to-item similarity measures for TV. Libby has written already about how we used these in ‘second screen’ (or ‘N-th’ screen, aka N-Screen) prototypes showing the impact that new Web standards might make on tired and outdated notions of “TV remote control”.

What if your remote control could personalise a view of some content collection? What if it could show you similar things based on your viewing behavior, and that of others? What if you could explore the ever-growing space of TV content using simple drag-and-drop metaphors, sending items to your TV or to your friends with simple tablet-based interfaces?

medieval_society

So that’s what we’ve been up to in NoTube. There are prototypes using BBC content (sadly not viewable by everyone due to rights restrictions), but also some experiments with TV materials from the Internet Archive, and some explorations that look at TED’s video collection as an example of Web-based content that (via ted.com and YouTube) are more generally viewable. Since every item in the BBC’s Archive is catalogued using a library-based classification system (Lonclass, itself based on UDC) the topic of cross-referencing books and TV has cropped up a few times.

new_colonialism

Meanwhile, in (the digital Public Library of) America, … the Harvard Library Innovation Lab team have a huge and fantastic dataset describing 14 million bibliographic records. I’m not sure exactly how many are ‘books'; libraries hold all kinds of objects these days. With the Harvard folk I’ve been trying to help figure out how we could cross-reference their records with other “Webby” sources, such as online video materials. Again using TED as an example, because it is high quality but with very different metadata from the library records. So we’ve been looking at various tricks and techniques that could help us associate book records with those. So for example, we can find tags for their videos on the TED site, but also on delicious, and on youtube. However taggers and librarians tend to describe things quite differently. Tags like “todo”, “inspirational”, “design”, “development” or “science” don’t help us pin-point the exact library shelf where a viewer might go to read more on the topic. Or conversely, they don’t help the library sites understand where within their online catalogues they could embed useful and engaging “related link” pointers off to TED.com or YouTube.

So we turned to other sources. Matching TED speaker names against Wikipedia allows us to find more information about many TED speakers. For example the Tim Berners-Lee entry, which in its Linked Data form helpfully tells us that this TED speaker is in the categories ‘Japan_Prize_laureates’, ‘English_inventors’, ‘1955_births’, ‘Internet_pioneers’. All good to know, but it’s hard to tell which categories tell us most about our speaker or video. At least now we’re in the Linked Data space, we can navigate around to Freebase, VIAF and a growing Web of data-sources. It should be possible at least to associate TimBL’s TED talks with library records for his book (so we annotate one bibliographic entry, from 14 million! …can’t we map areas, not items?).

tv

Can we do better? What if we also associated Tim’s two TED talk videos with other things in the library that had the same subject classifications or keywords as his book? What if we could build links between the two collections based not only on published authorship, but on topical information (tags, full text analysis of TED talk transcripts). Can we plan for a world where libraries have access not only to MARC records, but also full text of each of millions of books?

Screen%20shot%202011-10-11%20at%2010.15.07%20AM

I’ve been exploring some of these ideas with David Weinberger, Paul Deschner and Matt Phillips at Harvard, and in NoTube with Libby Miller, Vicky Buser and others.

edu

Yesterday I took the time to make some visual sanity check of the bibliographic data as processed into a ‘similarity space’ in some Mahout experiments. This is a messy first pass at everything, but I figured it is better to blog something and look for collaborations and feedback, than to chase perfection. For me, the big story is in linking TV materials to the gigantic back-story of context, discussion and debate curated by the world’s libraries. If we can imagine a view of our TV content catalogues, and our libraries, as visual maps, with items clustered by similarity, then NoTube has shown that we can build these into the smartphones and tablets that are increasingly being used as TV remote controls.

Screen%20shot%202011-10-11%20at%2010.12.25%20AM

And if the device you’re using to pause/play/stop or rewind your TV also has access to these vast archives as they open up as Linked Data (as well as GPS location data and your Facebook password), all kinds of possibilities arise for linked, annotated and fact-checked TV, as well as for showing a path for libraries to continue to serve as maps of the entertainment, intellectual and scientific terrain around us.

Screen%20shot%202011-10-11%20at%2010.16.46%20AM

A brief technical description. Everything you see here was made with Gephi, Mahout and experimental data from the Library Innovation Lab at Harvard, plus a few scripts to glue it all together.

Mahout was given 100,000 extracts from the Harvard collection. Just main and sub-title, a local ID, and a list of topical phrases (mostly drawn from Library of Congress Subject Headings, with some local extensions). I don’t do anything clever with these or their sub-structure or their library-documented inter-relationships. They are treated as atomic codes, and flattened into long pseudo-words such as ‘occupational_diseases_prevention_control’ or ‘french_literature_16th_century_history_and_criticism’,
‘motion_pictures_political_aspects’, ‘songs_high_voice_with_lute’, ‘dance_music_czechoslovakia’, ‘communism_and_culture_soviet_union’. All of human life is there.

David Weinberger has been calling this gigantic scope our problem of the ‘Taxonomy of Everything’, and the label fits. By mushing phrases into fake words, I get to re-use some Mahout tools and avoid writing code. The result is a matrix of 100,000 bibliographic entities, by 27684 unique topical codes. Initially I made the simple test of feeding this as input to Mahout’s K-Means clustering implementation. Manually inspecting the most popular topical codes for each cluster (both where k=12 to put all books in 12 clusters, or k=1000 for more fine-grained groupings), I was impressed by the initial results.

Screen%20shot%202011-10-11%20at%2010.22.37%20AM

I only have these in crude text-file form. See hv/_k1000.txt and hv/_twelve.txt (plus dictionary, see big file
_harv_dict.txt ).

For example, in the 1000-cluster version, we get: ‘medical_policy_united_states’, ‘health_care_reform_united_states’, ‘health_policy_united_states’, ‘medical_care_united_states’,
‘delivery_of_health_care_united_states’, ‘medical_economics_united_states’, ‘politics_united_states’, ‘health_services_accessibility_united_states’, ‘insurance_health_united_states’, ‘economics_medical_united_states’.

Or another cluster: ‘brain_physiology’, ‘biological_rhythms’, ‘oscillations’.

How about: ‘museums_collection_management’, ‘museums_history’, ‘archives’, ‘museums_acquisitions’, ‘collectors_and_collecting_history’?

Another, conceptually nearby (but that proximity isn’t visible through this simple clustering approach), ‘art_thefts’, ‘theft_from_museums’, ‘archaeological_thefts’, ‘art_museums’, ‘cultural_property_protection_law_and_legislation’, …

Ok, I am cherry picking. There is some nonsense in there too, but suprisingly little. And probably some associations that might cause offense. But it shows that the tooling is capable (by looking at book/topic associations) at picking out similarities that are significant. Maybe all of this is also available in LCSH SKOS form already, but I doubt it. (A side-goal here is to publish these clusters for re-use elsewhere…).

Screen%20shot%202011-10-11%20at%2010.23.22%20AM

So, what if we take this, and instead compute (a bit like we did in NoTube from ratings data) similarity measures between books?

Screen%20shot%202011-10-11%20at%2010.24.12%20AM

I tried that, without using much of Mahout’s sophistication. I used its ‘rowsimilarityjob’ facility and generated similarity measures for each book, then threw out most of the similarities except the top 5, later the top 3, from each book. From this point, I moved things over into the Gephi toolkit (“photoshop for graphs”), as I wanted to see how things looked.

Screen%20shot%202011-10-11%20at%2010.37.06%20AM

First results shown here. Nodes are books, links are strong similarity measures. Node labels are titles, or sometimes title + subtitle. Some (the black-background ones) use Gephi’s “modularity detection” analysis of the link graph. Others (white background) I imported the 1000 clusters from the earlier Mahout experiments. I tried various of the metrics in Gephi and mapped these to node size. This might fairly be called ‘playing around’ at this stage, but there is at least a pipeline from raw data (eventually Linked Data I hope) through Mahout to Gephi and some visual maps of literature.

1k_overview

What does all this show?

That if we can find a way to open up bibliographic datasets, there are solid opensource tools out there that can give new ways of exploring the items described in the data. That those tools (e.g. Mahout, Gephi) provide many different ways of computing similarity, clustering, and presenting. There is no single ‘right answer’ for how to present literature or TV archive content as a visual map, clustering “like with like”, or arranging neighbourhoods. And there is also no restriction that we must work dataset-by-dataset, either. Why not use what we know from movie/TV recommendations to arrange the similarity space for books? Or vice-versa?

I must emphasise (to return to Ben Fry’s opening remark) that this is a proof-of-concept. It shows some potential, but it is neither a user interface, nor particularly informative. Gephi as a tool for making such visualizations is powerful, but it too is not a viable interface for navigating TV content. However these tools do give us a glimpse of what is hidden in giant and dull-sounding databases, and some hints for how patterns extracted from these collections could help guide us through literature, TV or more.

Next steps? There are many things that could be tried; more than I could attempt. I’d like to get some variant of these 2D maps onto ipad/android tablets, loaded with TV content. I’d like to continue exploring the bridges between content (eg. TED) and library materials, on tablets and PCs. I’d like to look at Mahout’s “collocated terms” extraction tools in more details. These allow us to pull out recurring phrases (e.g. “Zero Sum”, “climate change”, “golden rule”, “high school”, “black holes” were found in TED transcripts). I’ve also tried extracting bi-gram phrases from book titles using the same utility. Such tools offer some prospect of bulk-creating links not just between single items in collections, but between neighbourhood regions in maps such as those shown here. The cross-links will never be perfect, but then what’s a little serendipity between friends?

As full text access to book data looms, and TV archives are finding their way online, we’ll need to find ways of combining user interface, bibliographic and data science skills if we’re really going to make the most of the treasures that are being shared in the Web. Since I’ve only fragments of each, I’m always drawn back to think of this in terms of collaborative work.

A few years ago, Netflix had the vision and cash to pretty much buy the attention of the entire machine learning community for a measly million dollars. Researchers love to have substantive datasets to work with, and the (now retracted) Netflix dataset is still widely sought after. Without a budget to match Netflix’, could we still somehow offer prizes to help get such attention directed towards analysis and exploitation of linked TV and library data? We could offer free access to the world’s literature via a global network of libraries? Except everyone gets that for free already. Maybe we don’t need prizes.

Nearby in the Web: NoTube N-Screen, Flickr slideshow

‘Republic of Letters’ in R / Custom Widgets for Second Screen TV navigation trails

As ever, I write one post that perhaps should’ve been two. This is about the use and linking of datasets that aid ‘second screen’ (smartphone, tablet) TV remotes, and it takes as a quick example a navigation widget and underlying dataset that show us how we might expect to navigate TV archives, in some future age when TV lives more fully in the World Wide Web. I argue that access to the ‘raw data‘ and frameworks for embedding visualisation apps are of equal importance when thinking about innovative ways of exploring the ever-growing archives. All of this comes from many discussions with my NoTube colleagues and other collaborators; rambling scribblyness is all my own.

Ben Hammersley points us at a lovely Flash visualization http://www.stanford.edu/group/toolingup/rplviz/”>Mapping the Republic of Letters”.

From the YouTube overview, “Researchers map thousands of letters exchanged in the 18th century’s “Republic of Letters” and learn at a glance what it once took a lifetime of study to comprehend.”


Mapping the Republic of Letters has at its center a multidimensional data set which spans 300 years and nearly 100,000 letters. We use computing tools that help us to measure and analyze data quantitatively, though that will not take us to our goal. While we use software and computing techniques that were designed for scientific and statistical methods, we are seeking to develop computing tools to enhance humanistic methods, to help us to explore qualitative aspects of the Republic of Letters. The subject of our study and the nature of the material require it. The collections of correspondence and records of travel from this period are incomplete. Of that incomplete material only a fraction has been digitized and is available to us. Making connections and resolving ambiguities in the data is something that can only be done with the help of computing, but cannot be done by computing alone. (from ‘methods and philosophy‘)


screenshot of Republic of Letters app, showing social network links superimposed on map of historical western Europe


See their detailed writeup for more on this fascinating and quite beautiful work. As I’m working lately on linking TV content more deeply into the Web, and on ‘second screen’ navigation, this struck me as just the kind of interface which it ought to be possible to re-use on a tablet PC to explore TV archives. Forgetting for the moment difficulties with Flash on iPads and so on, the idea roughly is that it would be great to embed such a visualization within a TV watching environment, such that when the ‘republic of letters’ widget is focussed on some person, place, or topic, we should have the opportunity to scan the available TV archives for related materials to show.

So a glance at Chrome’s ‘developer tools’ panel gave me a link to the underlying data used by the visualisation. I don’t know exactly whose it is, nor how they want it used, so please treat it with respect. Still, there it is, sat in the Web, in tab-separated format, begging to be used. There’s a lot you can do with the Flash application that I’ve barely touched, but I’m intrigued by the underlying dataset. In particular, where they have the string “Tonson, Jacob”, the data linker in me wants to see a Wikipedia or DBpedia link, since they provide explanation, context, related people, places and themes; all precious assets when trying to scrape together related TV materials to inform, educate or entertain someone with. From a few test searches, it turns out that (many? most?) the correspondents are quite easily matched to Wikipedia: William Congreve, Montagu, 1st earl of Halifax, CharlesHough, bishop of Worcester, John; Stanyan, Abraham;  … Voltaire and others. But what about the data?

Lately I’ve been learning just a little about R, a language used mainly for statistics and related analysis. Here’s what it’ll do ‘out of the box’, in untrained hands:

letters<-read.csv('data.txt',sep='\t', header=TRUE)
v_author = letters$Author=="Voltaire"
v_letters = letters[v_author, ]
Where were Voltaire’s letters sent?
> cbind(summary(v_letters$dest_country))
[,1]
Austria            2
Belgium            6
Canada             0
Denmark            0
England           26
France          1312
Germany           97
India              0
Ireland            0
Italy             68
Netherlands       22
Portugal           0
Russia             5
Scotland           0
Spain              1
Sweden             0
Switzerland      342
The Netherlands    1
Turkey             0
United States      0
Wales              0
As the overview and video in the ‘Republic of Letters‘ site points out (“Tracking 18th-century “social network” through letters”), the patterns of correspondence eg. between Voltaire and e.g. England, Scotland and Ireland jumps out of the data (and more so its visualisation). There are countless ways this information could be explored, presented, sliced-and-diced. Only a custom app can really make the most of it, and the Republic of Letters work goes a long way in that direction. They also note that
The requirements of our project are very much in sync with current work being done in the linked-data/ semantic web community and in the data visualization community, which is why collaboration with computer science has been critical to our project from the start.
So the raw data in the Web here is a simple table; while we could spend time arguing about whether it would better be expressed in JSON, XML or an RDF notation, I’d rather see some discussion around what we can do with this information. In particular, I’m intrigued by the possibilities of R alongside the data-linking habits that come with RDF. If anyone manages to tease anything interesting from this dataset, perhaps mixed in with DBpedia, do post your results.
And of course there are always other datasets to examine; for example see the Darwin correspondence archives, or the Open Knowledge Foundation’s Open Correspondence project which has a Dickens-based pilot. While it is wonderful having UI that is tuned to the particulars of some dataset, it is also great when we can re-use UI code to explore similarly structured data from elsewhere. On both the data side and the UI side, this is expensive, tough work to do well. My current concern is to maximise re-use of both UI and data for the particular circumstances of second-screen TV navigation, a scenario rarely a first priority for anyone!
My hope is that custom navigation widgets for this sort of data will be natural components of next-generation TV remote controls, and that TV archives (and other collections) will open up enough of their metadata to draw in (possibly paying) viewers. To achieve this, we need the raw data on both sides to be as connectable as possible, so that application authors can spend their time thinking about what their users really need and can use, rather than on whether they’ve got the ‘right’ Henry Newton.
If we get it right, there’s a central role for librarianship and archivists in curating the public, linked datasets that tell us about the people, places and topics that will allow us to make new navigation trails through Web-connected television, literature and encyclopedia content. And we’ll also see new roles for custom visualizations, once we figure out an embedding framework for TV widgets that lets them communicate with a display system, with other users in the same room or community, and that is designed for cross-referencing datasets that talk about the same entities, topics, places etc.
As I mentioned regarding Lonclass and UDC, collaboration around open shared data often takes place in a furtive atmosphere of guilt and uncertainty. Is it OK to point to the underlying data behind a fantastic visualisation? How can we make sure the hard work that goes into that data curation is acknowledged and rewarded, even while its results flow more freely around the Web, and end up in places (your TV remote!) that may never have been anticipated?

Local Video for Local People

OK it’s all Google stuff, but still good to see. Go to Google Maps, My Maps, to find ‘Videos from YouTube’ listed. Here’s where I used to live (Bristol UK) and where I live now (Amsterdam, The Netherlands). Here’s a promo film of some nearby art installations from ArtZuid, who even have a page in English. I wouldn’t have found the video or the nearby links except through the map overlay. I don’t know exactly how they’re geotagging the videos, I can’t see an option under ‘my videos’ in YouTube, so perhaps it’s automatic or viewer annotations. In YouTube, you can add a map link under ‘My Videos’ / ‘Edit Video'; I didn’t see that initially. I made some investigations into similar issues (videos on maps) while at Joost; see brief mention in my Fundamentos Web slides from a couple of years ago.
Oh, nearly forgot to mention: zooming out to get a Europe or World-wide view is quite striking too.

YouTube/Viacom privacy followup (and what Google should do)

A brief update on the YouTube/Viacom privacy disaster.

From Ellen Nakashima in the Washington Post:

Yesterday, lawyers for Google said they would not appeal the ruling. They sent Viacom a letter requesting that the company allow YouTube to redact user names and IP addresses from the data.

“We are pleased the court put some limits on discovery, including refusing to allow Viacom to access users’ private videos and our search technology,” Google senior litigation counsel Catherine Lacavera said in a statement. “We are disappointed the court granted Viacom’s overreaching demand for viewing history. We will ask Viacom to respect users’ privacy and allow us to anonymize the logs before producing them under the court’s order.”

I’m pleased to read that Google are trying to keep identifying information out of this (vast) dataset.

Viacom claim to want this data to “measure the popularity of copyrighted video against non-copyrighted video” (in the words of the Washington Post article; I don’t have a direct quote handy).

If that is the case, I suggest their needs could be met with a practical compromise. Google should make a public domain data dump summarising the (already public) favouriting history of each video (with or without reference to users, whose identifiers could be scrambled/obscured). This addresses directly the Viacom demand while sticking to the principle of relying on the public record to answer Viacom’s query. Only if the public record is incapable of answering Viacom’s (seemingly reasonable) request should users private behaviour logs be even considered. Google should also make use of their own Social Graph API to determine how many YouTube usernames are already associated in the public Web with other potentially identifying profile information; those usernames at least should not be handed over without at least some obfuscation.

If we know which YouTube videos are copyrighted (and Viacom owned). And we know how long they’ve been online, and which ones have been publicly flagged as ‘favourites’ by YouTube users, we have a massively rich dataset. I’d like to see that avenue of enquiry thoroughly exhausted before this goes any further.

Nearby in the Web: Danny Weitzner has blogged further thoughts on all this, including a pointer to a recent paper on information accountability, suggesting a possible shift of emphasis from who can access information, to the acceptable uses to which it may be put.

YouAndYouAndYouTube: Viacom, Privacy and the Social Graph API

From Wired via Thomas Roessler:

Google will have to turn over every record of every video watched by YouTube users, including users’ names and IP addresses, to Viacom, which is suing Google for allowing clips of its copyright videos to appear on YouTube, a judge ruled Wednesday.

I hope nobody thought their behaviour on youtube.com was a private matter between them and Google.

The Judge’s ruling (pdf) is interesting to read (ok, to skim). As the Wired article says,

The judge also turned Google’s own defense of its data retention policies — that IP addresses of computers aren’t personally revealing in and of themselves, against it to justify the log dump.

Here’s an excerpt. Note that there is also a claim that youtube account IDs aren’t personally identifying.

Defendants argue that the data should not be disclosed because of the users’ privacy concerns, saying that “Plaintiffs would likely be able to determine the viewing and video uploading habits of YouTube’s users based on the user’s login ID and the user’s IP address” .

But defendants cite no authority barring them from disclosing such information in civil discovery proceedings, and their privacy concerns are speculative.  Defendants do not refute that the “login ID is an anonymous pseudonym that users create for themselves when they sign up with YouTube” which without more “cannot identify specific individuals”, and Google has elsewhere stated:

“We . . . are strong supporters of the idea that data protection laws should apply to any data  that could identify you.  The reality is though that in most cases, an IP address without additional information cannot.” — Google Software Engineer Alma Whitten, Are IP addresses personal?, GOOGLE PUBLIC POLICY BLOG (Feb. 22, 2008)

So forget the IP address part for now.

Since early this year, Google have been operating an experimental service called the Social Graph API. From their own introduction to the technology:

With so many websites to join, users must decide where to invest significant time in adding their same connections over and over. For developers, this means it is difficult to build successful web applications that hinge upon a critical mass of users for content and interaction. With the Social Graph API, developers can now utilize public connections their users have already created in other web services. It makes information about public connections between people easily available and useful.

Only public data. The API returns web addresses of public pages and publicly declared connections between them. The API cannot access non-public information, such as private profile pages or websites accessible to a limited group of friends.

Google’s Social Graph API makes easier something that was already possible: using XFN and FOAF markup from the public Web to associate more personal information with YouTube accounts. This makes information that was already public increasingly accessible to automated processing. If I choose to link to my YouTube profile with the XFN markup rel=’me’ from another of my profiles,  those 8 characters are sufficient to bridge my allegedly anonymous YouTube ID with arbitrary other personal information. This is done in a machine-readable manner, one that Google has already demonstrated a planet-wide index for.

Here is the data returned by Google’s Social Graph API when asking for everything about my YouTube URL:

{
 "canonical_mapping": {
  "http://youtube.com/user/modanbri": "http://youtube.com/user/modanbri"
 },
 "nodes": {
  "http://youtube.com/user/modanbri": {
   "attributes": {
    "url": "http://youtube.com/user/modanbri",
    "profile": "http://youtube.com/user/modanbri",
    "rss": "http://youtube.com/rss/user/modanbri/videos.rss"
   },
   "claimed_nodes": [
   ],
   "unverified_claiming_nodes": [
    "http://friendfeed.com/danbri",
    "http://www.mybloglog.com/buzz/members/danbri"
   ],
   "nodes_referenced": {
   },
   "nodes_referenced_by": {
    "http://friendfeed.com/danbri": {
     "types": [
      "me"
     ]
    },
    "http://guttertec.swurl.com/friends": {
     "types": [
      "friend"
     ]
    },
    "http://www.mybloglog.com/buzz/members/danbri": {
     "types": [
      "me"
     ]
    }
   }
  }
 }
}

You can see here that the SGAPI, built on top of Google’s Web crawl of public pages, has picked out the connection to my FriendFeed (see FOAF file) and MyBlogLog (see FOAF file) accounts, both of whom export XFN and FOAF descriptions of my relationship to this YouTube account, linking it up with various other sites and profiles I’m publicly associated with.

YouTube users who have linked their YouTube account URLs from other social Web sites (something sites like FriendFeed and MyBlogLog actively encourage), are no longer anonymous on YouTube. This is their choice. It can give them a mechanism for sharing ‘favourited’ videos with a wide circle of friends, without those friends needing logins on YouTube or other Google services. This clearly has business value for YouTube and similar ‘social video’ services, as well as for users and Social Web aggregators.

Given such a trend towards increased cross-site profile linkage, it is unfortunate to read that YouTube identifiers are being presented as essentially anonymous IDs: this is clearly not the case. If you know my YouTube ID ‘modanbri’ you can quite easily find out a lot more about me, and certainly enough to find out with strong probability my real world identity. As I say, this is my conscious choice as a YouTube user; had I wanted to be (more) anonymous, I would have behaved differently. To understand YouTube IDs as being anonymous accounts is to radically misunderstand the nature of the modern Web.

Although it wouldn’t protect against all analysis, I hope the user IDs are at least scrambled before being handed over to Viacom. This would make it harder for them to be used to look up other data via (amongst other things) Google’s own YouTube and Social Graph APIs.

Update: I should note also that the bridging of YouTube IDs with other profiles is one that is not solely under the control of the YouTube user. Friends, contacts, followers and fans on other sites can link to YouTube profiles freely; this can be enough to compromise an otherwise anonymous account. Increasingly, these links are machine-processable; a trend I’ve previously argued is (for better or worse) inevitable.

Furthermore, the hypertext and data environment around YouTube and the Social Web is rapidly evolving; the lookups and associations we’ll be able to make in 1-2 years will outstrip what is possible today. It only takes a single hyperlink to reveal the owner of a YouTube account name; many such links will be created in the months to come.

Google Data APIs (and partial YouTube) supporting OAuth

Building on last month’s announcement of OAuth for the Google Contacts API, this from Wei on the oauth list:

Just want to let you know that we officially support OAuth for all Google Data APIs.

See blog post:

You’ll now be able to use standard OAuth libraries to write code that authenticates users to any of the Google Data APIs, such as Google Calendar Data API, Blogger Data API, Picasa Web Albums Data API, or Google Contacts Data API. This should reduce the amount of duplicate code that you need to write, and make it easier for you to write applications and tools that work with a variety of services from multiple providers. [...]

There’s also a footnote, “* OAuth also currently works for YouTube accounts that are linked to a Google Account when using the YouTube Data API.”

See the documentation for more details.

On the YouTube front, I have no idea what % of their accounts are linked to Google; lots I guess. Some interesting parts of the YouTube API: retrieve user profiles, access/edit contacts, find videos uploaded by a particular user or favourited by them plus of course per-video metadata (categories, keywords, tags, etc). There’s a lot you could do with this, in particular it should be possible to find out more about a user by looking at the metadata for the videos they favourite.

Evidence-based profiles are often better than those that are merely asserted, without being grounded in real activity. The list of people I actively exchange mail or IM with is more interesting to me than the list of people I’ve added on Facebook or Orkut; the same applies with profiles versus tag-harvesting. This is why the combination of last.fm’s knowledge of my music listening behaviour with the BBC’s categorisation of MusicBrainz artist IDs is more interesting than asking me to type my ‘favourite band’ into a box. Finding out which bands I’ve friended on MySpace would also be a nice piece of evidence to throw into that mix (and possible, since MusicBrainz also notes MySpace URIs).

So what do these profiles look like? The YouTube ‘retrieve a profile‘ API documentation has an example. It’s Atom-encoded, and beyond the video stuff mentioned above has fields like:

  <yt:age>33</yt:age>
  <yt:username>andyland74</yt:username>
  <yt:books>Catch-22</yt:books>
  <yt:gender>m</yt:gender>
  <yt:company>Google</yt:company>
  <yt:hobbies>Testing YouTube APIs</yt:hobbies>
  <yt:location>US</yt:location>
  <yt:movies>Aqua Teen Hungerforce</yt:movies>
  <yt:music>Elliott Smith</yt:music>
  <yt:occupation>Technical Writer</yt:occupation>
  <yt:school>University of North Carolina</yt:school>
  <media:thumbnail url='http://i.ytimg.com/vi/YFbSxcdOL-w/default.jpg'/>
  <yt:statistics viewCount='9' videoWatchCount='21' subscriberCount='1'
    lastWebAccess='2008-02-25T16:03:38.000-08:00'/>

Not a million miles away from the OpenSocial schema I was looking at yesterday, btw.

I haven’t yet found where it says what I can and can’t do with this information…

WHY MIGHT CONNECTING WITH ZANDER JULES BE A GOOD IDEA?

Or: towards evidence-based ‘add a contact’ filtering…

This just in from LinkedIn:

Have a question? Zander Jules’s network will probably have an answer
You can use LinkedIn Answers to distribute your professional questions to Zander Jules and your extended network. You can get high-quality answers from experienced professionals.

Zander Jules requested to add you as a connection on LinkedIn:

Dan,

Dear
My name is Zander Jules a Banker and accountant with Bank Atlantique Cote Ivoire.I contacting u for a business transfer of a large sum of money from a dormant account. Though I know that a transaction of this magnitude will make any one apprehensive,
but I am assuring u all will be well at the end of the day.I am the personal accounts manager to Engr Frank Thompson, a National of ur country, who used to work with an oil servicing company here in Cote Ivoire. My client, his wife & their 3 children were involved in the ill fated Kenya Airways crash in the coasts of Abidjan in January 2000 in which all passengers on board died. Since then I have made several inquiries to ur embassy to locate any of my clients extended relatives but has been unsuccessful.After several attempts, I decided to trace his last name via internet,to see if I could locate any member of his
family hence I contacted u.Of particular interest is a huge deposit with our bank in our country,where the deceased has an account valued at about $16 million USD.They have issued me notice to provide the next of kin or our bank will declare the account unservisable and thereby send the funds to the bank treasury.Since I have been unsuccessful in locating the relatives for past 7 yrs now, I will seek ur consent to present you as the next of kin of the deceased since u have the same last names, so that the proceeds of this account valued at $16million USD can be paid to u and then u and I can share the money.All I require is your honest cooperation to enable us see this deal through. I guarantee that this will be executed under all legitimate arrangement that will protect you from any breach of the law. In your reply mail, I want you to give me your full names, address, D.O.B, tel& fax #.If you can handle this with me, reach me for more details.

Thanking u for ur coperation.
Regards,

I’m suprised we’ve not seen more of this, and sooner. Youtube contacts are pretty spammy, and twitter have also suffered. The other networks are relatively OK so far. But I don’t think they’re anything like as robust as they’ll need to get, particularly since a faked contact can get privileged access to personal details. Definitely an arms race…

Videos from SemanticCamp Paris 1

Spotted on the websemantique list, a Youtube playlist of videos from SemanticCamp Paris 1,  16 février. I think they’ve just had a second SemanticCamp already, but these five videos are from the earlier event. Lots of FOAF, RDFa etc.

 L’objectif du SemanticCamp Paris est de créer les conditions pour que les développeurs, les étudiants, les managers et les chercheurs se rencontrent sur le thème du Web Sémantique.