Mathew Lowry

7/10/13: I received a couple of interesting reactions to the post about rebooting BloggingPortal, but as some of them were by email I decided to reply in FAQAO (Frequently Asked Questions And Objections) format, which I just invented, in case others have the same questions. If I get more I’ll probably just edit this post, with each re-edition highlighting new Q&Os.
Latest addition: 13/10/13.

So far:

 

You don’t really expect volunteers to step forward, do you?

On “curation” I have to wonder what the point is: To actually manually aggregate posts? You end up getting a random splurge … depending on whether people feel like aggregating that day. (comment by CraigJWilly)

Curated sections? in the past … we did not even manage to organise a simple tagging exercise – so where do all those super motivated volunteers come from? Again, it is not realistic. (by email)

I’m glad I started posting so early in this process, because these questions show how poorly I’ve explained what ‘machine semantic analysis’ can actually do for the EU online public space.

The whole reason BP stopped functioning is it relied on volunteers manually tagging and highlighting posts. What a lot of BP editors seemed to have missed from my original 2009 proposals is that semantic analysis software can read and tag any document automatically, using a coherent taxonomy.

Hence, from my post (seriously, doesn’t anyone read?):

“if the volunteers don’t come back, and are not replaced, then at least this engine will chug away, automatically translating and classifying posts…”

I have used this tech a number of times over the past 4 years and have posted about it) a few times. The technology’s not perfect, but it currently works OK, it’ll only get better and it certainly beats non-existent volunteer labour. The misuse of this technology by (mainly) US media companies may make it controversial (see What “The Filter Bubble” means for the Brussels Bubble), but this is just basic curation we’re talking here.

So, NO, I don’t expect this to run off volunteer labour. You think I’m an idiot? No, don’t answer that.

 

What about national or thematic categories?

“What might be useful: … organized sections (good taxonomy/categories) automatically aggregating posts from blogs by category (country; topic; profession of blogger…). The added value would be in bringing together the major national blogs on that topic in the different countries.” (CraigJWilly)

That’s exactly the goal. Minus classifying bloggers – we’re expanding the scope beyond bloggers (see below), and I can’t see how that could be done with any accuracy anyway.

Again, this is totally automatic:

  • The semantic analysis software tags the content using a number of taxonomies, including country, which gives the geographic sections you want to see, but more granular – we’ll classify posts, not entire blogs, by country.
  • Other taxonomies classify posts by topic, while machine language detection would classify each post by original language.

And because there’ll also be faceted search (think of any catalogue-based site, or read wikipedia), you’ll be able to combine these tags together.

That means users could drill down to, say, posts about “EU agricultural policy AND France AND written in English”, simply by clicking on the ‘agriculture’ tag and then adding ‘related tags’ (i.e., ‘France’, ‘in EN’) to continuously refine your search. Probably to find something nasty by the Economist (again, we’re not limiting ourselves to blog posts).

These tags can also be used to provide a ‘recommended reading’ feature – i.e., “if you’re interested in that, you may be interested in this too”.

And while I’m on the topic of search, we’ll be indexing the entirety of each post, in both original language and machine-EN-translation, so users will be able to free-text search the EU online public space much more effectively via BP than anywhere else.
 

How will this add value anyway?

content discovery is happening on twitter, facebook, reddit, google+ or whatever comes next – also online socialising tends to happens there (or even in forums..) and not on thematic platforms. google is also a pretty good search engine – so what is the added value of a new bp? (by email)

Good question. I believe the above combination of machine translation and tagging plus faceted search, all focused on one niche, will do a better job than Google in helping people find opinions published on (EU) policy, particularly in other languages.

How often have you used Google to find a blog post on a topic written in, say, Bulgarian? BP will apply semantic technology to the autotranslation of each post. Does Google, let alone thematic platforms, do that?

A better question is to ask who this will add value for. The beneficiaries are anyone interested in having a conversation on EU policy:

  • bloggers, both to build their network and audience, and to understand the conversation going on in other Member States and/or in the Brussels Bubble
  • policymakers and journalists, researching views and opinions for the same reason
  • NGOs, lobbyists & communications teams, outreaching to bloggers and social media spaces to raise awareness of an issue, policy, programme, etc.

In other words, all denizens of the EU online public space. Because they all need to do one thing first: Listen to What People are Saying. Which is pretty hard to when you don’t know where they are.

 

Why bother? Blogging’s dead!

“blogs are dead, or have gone so mainstream that it is not worth thinking about it as one format anymore …if you are ambitious and you want to write about politics you would not even think of opening a blog in 2013” (by email)

The death of blogging is one of the oldest stories of the millennium – I first heard that blogging was dead in maybe 2006, Fast Company reiterated it last December, and another article (usually a blog post) for and against appears every day. Maybe it feels that way because blogging used to be the only thing there was in online communities. Now it’s the slightly old fashioned uncle at the Christmas party where everyone’s celebrating the latest social media platform.

There are two answers to this question w.r.t. BloggingPortal:

When it comes to actually discussing – as opposed to liking or promoting – something as weighty as an EU policy, the long form blog still has a role to play. Blogging’s on the back foot because it’s hard to build up a blogging audience in a noisy world, and much easier to get a quick fix on Facebook. By curating posts, Bloggingportal will help alleviate that, just as BlogActiv’s launch stimulated blogging in Brussels by providing instant audience.

Before you write off blogging, ask yourself:

  • when was the last time you had a thoughtful exchange on Facebook or Twitter on EU policy? Where you actually learnt something new? Or taught somebody something?
  • if blogging’s dead, why did Facebook, Twitter and Google+ all recently launch embeddable posts?*
  • why are over 358 million people viewing more than 11.3 billion blog posts each month on WordPress, where users produce about 47.2 million new posts and 68.7 million new comments each month?*
  • why are new blogging systems – Medium, Ghost, etc. – appearing all the time?

* hat tip: 3 Reasons That Blogging Isn’t Dead

The second answer is that while blogging ain’t dead, it is evolving fast. Which brings us to the next FAQAO:
 

Why call it bloggingportal?

“Do not call it bloggingportal” (CraigJWilly)

I’ve never liked the name ‘bloggingportal’, because it’s too media-specific. The idea is to be a discovery engine for opinion on EU policy – whether it be blog posts, newspaper articles, wikis, whatever.

Basically, if it’s about EU policy, and it has a stream that we can curate and process, it should go into Bloggingportal. I don’t really care whether it’s technically a blog or not.

So a better name would be … better. I’m not losing sleep over the name right now, but all name suggestions gratefully received. We could run a naming competition, or at least a poll.

 

Who runs Bloggingportal today, and How?

“Who keeps the lights on? Do any editors work at the institutions? How is the site run at the moment?”– by email

There are some FAQs on the site, but they are not being maintained – whereas the (automatically updated) home page today mentions that the site curates over 1100 blogs, for example, the FAQ hasn’t been updated since the total was around 750.

So the list of editors is probably out of date, too. I say ‘probably’ because in truth nobody really knows – people became editors by asking; the only times anyone was refused was because they were from an EU Institution and we wanted to avoid obvious conflicts of interest.

There is, in short, no central management or organisation at all. One guy owns the domain name, the other the server (I only launched this reboot process when they both agreed to pass these assets over to a to-be-created organisation if and when we get that far), and Editors curate posts and tweet in BP’s name as they feel like it.

This agreeable anarchy was fine and dandy until we started to discuss overhauling the site back in 2009. As I mentioned in Happy Birthday, BloggingPortal(?), the complete lack of structure meant that everyone had a veto. I had One Last Crack at it late last year, the veto-ers relented and we have a process allowing us to move forward – see How will it be run?, below.

The only thing we have tying it together is an Editors mailing list. If one tenth of the effort put into the emails flowing through that list had been focused and coordinated, we’d have relaunched Bloggingportal years ago.

 

I am still confused

“All “blog platforms” I know have not fulfilled their promise and largely disappeared again – so why should it be different for bp?”– by email

Perhaps because BP is not a blogging platform? It is a curation engine. It curates content published elsewhere, helping you find it by language, country and topic.

Blogging platforms are where some of that content is published. And they’re not dead – I launched the one you’re looking at now 7 years ago, and it’s going fine thanks very much.

(Well, actually, no, it’s showing its age. But it hasn’t disappeared, and with around 30 posts reaching the home page every few days, doesn’t exactly seem lifeless.)
 

What about social signals?

“For each section (topic, as above, or language), presentation of the latest posts could be based on “liking” or “+1ing” a post. That way you’d have interactivity, it could be community-driven. Just an idea.” (CraigJWilly)

Sure. The current spec includes:

  • using the ‘social media visibility’ of each piece of content to drive the interface (e.g., “most liked content”). We’ve got some ideas of how (e.g., grabbing data from the AddThis plugin, using the platforms’ APS…), but we’re not at that point in the process yet (see below)
  • adding a Reddit-style upvoting system – frankly I doubt we’ll get that level of engagement, so it’s in the spec as a Nice2Have for the moment.

 

Why not use tool X / Y / Z?

“Maybe we just need something like fuego plus a new fancy looking website”– by email

Sure. One of the reasons I’m doing this with maximal transparency is to harvest ideas like that, because there are probably 100 great tools out there. Fuego’s one. Rebelmouse also agreed to provide some premium services for free.

But we haven’t got to the technology choice stage yet. Instead, we’re following a CMS Choice Process, developed The Hard Way many years ago. It goes like this:

  • First figure out what you want – draw a picture of perfection, sans compromis
  • Prioritise the features
  • Do a wide and shallow survey of all available tools against the top priority features, and draw up a shortlist
  • Do a narrow and deep evaluation of the shortlisted tools using against all of the features

Of course, it helps to know before you start that what you want can be done by at least one technology, without spending a fortune. And they can – I’ve used the semantic and faceted search tools in Drupal-driven sites for the European Commission a number of times over the past 4-5 years, so the current specs reflect that.

I’ll try to make the next version, to be used to get some ballpark budget figures, more tech-agnostic, although the wireframes will probably betray my background with Drupal. Can’t be helped.

 

How will it be run?

” It was be ruled with an “iron fist” with perfect coincidence of decision-making and responsibility with a tight executive (1, 2 or 3 people). Not diffuse “consensus” (collective irresponsibility) or “rule by mailing list.””– CraigJWilly

Whether you meant “must” be ruled by an iron fist or not, I agree with the gist. As you know, I was ready to move ahead on this in 2009, but a veto by (in the end) just one person left us immobile for over 3 years.

But we cut that Gordian knot last winter and now have a process, while the people who own the database and the domain have agreed to pass them over to a non-profit organisation if and when it’s ready.

But that will only be set up if we ever get to the point where we actively start soliciting funds and support. We’re nowhere near that point yet. Right now, this is at best a two-man show a one-man show.

Tweet about this on TwitterShare on Facebook0Share on Google+1Share on LinkedIn1
Author :
Print

Comments

  1. A very detailed outline which gives indeed a good picture of where you are heading.

    One question on the automatic tagging (because this is the key concept I think): have the OpenCalais and Alchemy (I think these are the only two relevant ones) algorithms become better over the last 1-2 years? Because the last time I tried them it was kind of ok in English, but close to unusable for example in German.

    One other challenge is to filter EU-related content from bigger streams, for example CNN, or big nationals newspapers.

  2. The return of the BrusselsBlogger! It’s been too long since your last post … asking for help for Bloggingportal, 18 months ago.

    I’ve used OpenCalais a few times and it worked well, but only processing English text.

    My starting point is to machine-translate everything into English and then semantically process that, because that way we’ll have a single (albeit English only) taxonomy. The actual taxonomy can be manually translated for the other language interfaces (at least in Drupal it can).

    However I haven’t actually looked into automatic semantic analysis tools in any depth for a couple of years, and I know it’s an area of frenetic research and development – in fact I only mention OpenCalais because I’ve used the Drupal module, so I know that what I have in mind is feasible, even easy.

    But I’d really like to talk to some developers who are more up to date.

    Anyone know any?

  3. I was actually thinking the same: translating before analysing might be more efficient. Most of these systems like Open Calais and Alchemy are only available in a few languages anyway.

    Another point I want to raise is scaling the system. Already the current bloggingportal architecture can hardly deal with the 1000+ sources in a timely manner. If you add extraction of full text + counting of social media shares (which is nearly a real-time thing) and content analysis to it and still want it to deal with thousands of more sources then the technical architecture becomes a challenge on its own.

    A fascinating article on the technical setup for a similar system built by Prismatic (not uninteresting in the Bloggingportal context by the way) can be found here: http://highscalability.com/blog/2012/7/30/prismatic-architecture-using-machine-learning-on-social-netw.html

Leave a Reply