Report from NILM2014@London on building an online discussion venue and wiki for NILM


The first "NILM in London" workshop was held on Wednesday 3rd September. It was a lot of fun and we had some great conversations. In this blog post, I'd like to try to summarise the discussion around building an online discussion forum for NILM. There was strong support for having an online venue (e.g. an email list) to discuss NILM. Use cases cited included helping folks to stay up to date with new papers, workshops, software releases; to discuss technical challenges etc.

How 'open' should the community be?

If I remember correctly, everyone at the workshop agreed that mailing list content should be readable by everyone.

Everyone also appeared to agree that a risk with online communities is that they can become noisy and/or high-traffic and so require some level of moderation. We didn't discuss this in much detail.

I wrote a blog post almost a year ago with a proposal to start an online community for NILM. Why hasn't it happened yet? The honest answer is that I'm afraid it will guzzle up too much time with noisy questions. But, after seeing the enthusiasm for an online community, it definitely looks like we should go ahead and do it.

Which platform?

I think we agreed that we should first look at Google Groups. If that isn't sufficient for our needs then we should look elsewhere.

After the event, I had a quick look at the permissions settings on the Google Groups support pages. It seems that they have sufficient controls.


Please let me know if this plan sounds OK (I've also added some specific questions...)

  • we set up a Google Group called "NILM" (or "energy disaggregation"? or something else? Or should we use a different platform? Google+ communities?)
  • new users can subscribe instantly, without requiring moderation (we had discussed the idea that new users would have to request entry to the group and then moderators would accept / reject the request. But when would we ever reject a request? There just isn't enough information available at this point to judge. We can always change the permissions later.)
  • a user's first few posts will be moderated. Once they get into the rhythm of the group then they will be able to post immediately (without moderation).
  • One big question is whether to have 'sub topics'. e.g. we could have sub topics on 'New research papers', 'upcoming events', 'NILM algorithms', 'datasets', 'validation', 'jobs' etc. I don't think we discussed this at all on Wednesday. My hunch is that we should start with a single discussion forum (i.e. no sub topics). Maybe the community is sufficiently small to mean that we don't need to subdivide the forum? Later, if we need to, we can subdivide the forum. Also, if we start with a single forum then this will also help to clarify which subforums are required!

Does that sound OK?


We also briefly discussed building a NILM wiki (building a NILM wiki is something I blogged about almost a year ago. The general feeling on the NILM workshop on Wednesday was that I should just get on with building it and present it at the next meetup! At the NILMTK workshop on Thursday 4th, we discussed how useful it would be to use the wiki list details and quirks of NILM datasets.



I wonder if we should start with two 'sub-communities':

  • announcements (new papers, NILM workshops, software releases, jobs, new datasets etc). This feels like it will basically just be a bunch of links to other web pages. So maybe a mailing list isn't the best format for 'announements'? Maybe it should be a Google + community? Or a social bookmarking site like Reddit?
  • discussion (discussion of new papers; 'how do I do X?', 'let's discuss NILM validation' etc)

The reason for having two subcommunities is that some people will only want announcements without lots of noisy discussion.


Any thoughts?

I don't think that there is a problem with the 2 communities and it makes sense. As more people participate, I think that it will become clearer when new sub-groups should be created.

I'd be inclined to keep them all on the same platform (be it Google + , mailing lists etc.) since part of the point is to have it all in one central place?


Just wanted to move this topic forward again a little. We have been playing around with some and have some info looking into pricing.

There is one that we can embed into a website pretty easily to use and free, though will go up to about $90 per month if we want to have full user access rights.

The one that we use for our LabVIEW user groups are already paid for by national instruments and built off the technology from jivesoft.

They are pretty good and have all the poll features etc.

we are getting some pricing back on that one too.

Google groups are good for an email only discussion but can get messy quickly after a few replies and not great for separating topics.

I have got our marketing team looking into it further.

Hope all's well!


Hi Peter,

Thanks loads for looking into this.

Perhaps some desirable features of any NILM discussion platform are:

  • free (even if we can afford to pay for it for the next, say, 12 months - what happens in the long term? I don't think we want a community that's going to die if we all move on to a different research topic in 5 year's time)
  • as simple as possible to use and/or as familiar as possible. People just don't have the time to invest in figuring out new software.
  • hosted. We don't have time to keep the software updated ;)
  • likely to be around in 10 year's time

I'd be very tempted to advocate something like Google Groups or Google Plus. Google Groups does have a number of alternative "types" so can be set up to run very much like a traditional web forum if you want. If you don't have your email organised by thread then, I agree, any mailing list discussion can get messy in your inbox... but that's fixable by looking at the discussion on the web (where they are organised by thread). Google Groups also allows for tags and/or categories to be assigned to threads, to further help with organisation.

I also suspect that the NILM community is so small that any NILM mailing list will be fairly low-traffic (especially if we're successful in filtering out all the "please write my BSc research project literature review for me" type questions that I'm sure we've all received into our email inboxes over the years).

A calendar might be useful but that's easy to do with a shared iCal calendar or using to help organise future NILM Europe meetups (which I'd advocate) or a wiki to keep a list of events etc...

Polls might also be useful every now and then but we could just use Doodle for polls.

Thanks, Jack

Hey Guys,

I totally agree with the four points from above

I think hosting it is the key one, some guys in America have brought and etc are also gone.

We have bought these domains:


I think the three of us should access to the main admin account so that it has a driving force, then we can pass it on when required if either of you guys leave your post/industry or we get brought etc. We can buy the hosting for 10 years for cheap and if we switch it to godaddy it will be free.

Simple to use as possible, agreed and goes as standard.

hosted, agreed again as above.

likely to be around in 10years. - this needs to happen but will only really become the place we want it to if it is looked after and maintained?

Am off to a meeting this afternoon but can have a discussion tomorrow if that works, would be good to keep the ball rolling on this.

Best regards Peter

On the topic of domain names... a year ago, when we first started discussing a NILM wiki and discussion platform, we threw around some ideas for domain names and voted on our favorites. The full list of domain names we considered is available here. I registered these four domain names (and still have them):


I think our original idea was to install an instance of (Semantic) MediaWiki on my Site5 web host, which would be accessible from the above URLs. And then have a link from the front page of the wiki to the discussion forum (which could be a Google Groups thing or an alternative platform if we wanted). Does that still sound like a viable plan?

(I doubt we've got the time to try to closely integrate the wiki with the discussion forum. I tried doing this several years ago and it was a PITA. Maybe it's gotten easier. Not sure.)

(I won't have time to do the wiki until I finish at IBM. i.e. it won't be done until November 2014 at the earliest. But that needn't stop us from setting up a discussion forum and just putting a link on all of the above URLs to forward to the discussion forum page.)

I think the Google options make the most sense for now and we can forward from the domain names, though for simplicity, I think that we should just choose one domain name, otherwise, guaranteed that everyone will reference a different one. is the shortest? (PS. Am not really bothered either way if someone feels more strongly :-))

@Peter (and others): any objections if we go ahead and set up a Google group? No worries either way ;)

Hey Jack,

Is the shortest one not

or is also shorter no?


Best regards


Domain Name

I think it would be nice if the mailing list had global scope because there are NILM researchers in the US, India, Canada, Japan, China etc. So I'd be hesitant to go with any name which limits the scope to Europe. When Europe-centric issues arrise (like organising an event in Europe) then we could just do that in a single thread of start a new sub-forum or use tags or something.

Also, I have a hunch that people don't bother remembering domain suffixes any more. Intead they just Google for the main part of the domain name! Kind of like using Google as a contact book for URLs.

'' is short but it's also pretty non-descriptive unless your already know what the site is about. You can disaggregate all sorts of things (earth, audio, finances etc etc). So I'd personally go with something that's self descriptive to even people who are new to the field. As such, I'd personally lean towards '' but I'm really not that bothered! I do agree with Jim that we should pick one though (instead of using a whole set of URLs for the same site).

Mailing list

Should we go ahead and set up a Google Groups mailing list (that would be linked to from which ever domain name we pick? And then, further down the line, could also host our wiki).

@Peter, should I go ahead and set up a Google Groups NILM mailing list?

And is '' and OK domain name? (see above for a more detailed discussion)

I finish at IBM on Fri 24th Oct so we could meet up some time after that if you'd like to meet in person to discuss it.

We could perhaps use a Google Groups structure like the ML-News google group.

After meeting up with Peter to discuss this in more detail, I think we're converging on a plan (Peter, please correct me if I'm wrong!)...

Use as a place for displaying static information about our European NILM meetups (the first of which happened in London on the 3rd Sept). will behave a little like the website.

Then we'll set up a Google Groups mailing list for (global) discussion of NILM issues. We'll link to this mailing list from (in the not too distant future, should also host a NILM wiki).

The Google Groups mailing list will be modelled loosely on the ML-News google group.

Perhaps we could kick off the NILM mailing list with the following topics:

  • Technical NILM discussions: theory & software / hardware development (for example, this could be where we discuss how to design and organise a NILM competition / validation tool)
  • Events (meetups, workshops, conferences etc) (This would be used for announcing new events and also for discussing the organisation of events)
  • Announcement and discussion of NILM papers and patents
  • Calls for papers
  • Jobs and academic positions
  • General discussion

Does that sound OK???

Add new comment