A little attention.

Both KrankyChloe and Ohlaylala have both received a little good will and attention from 2 esteemed bloggers in the 'demographic' industry.

John Griffiths, pushing for hyper-localization.

Kyle Studstill, welcoming normalization.

Both issues are on the agenda.

Meanwhile, verysexyseo, is drawing quite a following and receiving, what I would consider, a large amount of DMs. 25 per day. Quite a lot isn't it? Many of them are solutions to 'making more friends'. They just keep pouring in. All the time.

xibipiio

The Siblings have an interesting order of existence - technically they don't exist until they do something - such as post a message. For an audience (watching) the experience is either fleeting or persisting though imagination.

James Bridle told me about Daniel Everertt over brunch today - whilst discussing the DGRs. Everett spent a slab of time in with the Pirahã People of Brazil (The Hi'aiti'ihi) as a Christian Missionary - charged with slowly explaining Jesus & co to the tribes.


Map from spiegel.de Article on the adventures of Everett.

In short, they converted him to their way of seeing - which is based upon 3 simple rules.

1. Did you experience, what you are telling me, first hand?
2. Did you get the information from a friend?
3. Did you deduce this information?

As such, they hold the idea that experience is everything - to which bleeds into their habits. The name of Everett's account of the tribes, is "Don't Sleep, There are Snakes." is the functional, meaningful way of saying "Good night" - which is a hollow western gesture. Further, because of the snakes, the Pirahã People nap during the day and night, for 15 minutes to 2 hours. Ah, Tribe life.

There's loads of good assumptions debunked in Everett's writing - how language and state of being are not essentially routed in mythology.

Here he is talking at the Long Now Foundation. He explains the term xibipiio, which is about 'slipping in and out of experience', a kin to 'peek-a-boo' games children play. Fascinating stuff this for language rules and 'game play'.



And a BBC Radio 4 reading from Everett. All good stuff. Plenty to borrow and learn from for the Siblings 'development'. The lack of a creation myth and fictionals in their culture forces them to exist only in the present. I like this as an approach to the DGR Siblings existence - they don't need faith - they only need existence.



In terms of 'arts', Wikipedia notes:

They have very little artwork. The artwork that is present, mostly necklaces and drawn stick-figures, is used primarily to ward off evil spirits.

If the Siblings are to have creativity, then I have to ask what would be the evil spirits within their digital dimension?


Further reading to be had everywhere on the web, but this Times article is a good overview.

Here's the Amazon link for the book - "Don't Sleep, There are Snakes."

Thanks James!

Picturing Thoughts

So as the Siblings RTweet away, it's nice to wonder what images would be on their mind.

Using a classique Twitter & Flickr mashup via Keyword Extraction using Yahoo!Pipes, you get a slideshow based upon their Retweets.

Here's Jimmmo's 'mind' or 'memory'. Whatever you want to perceive it as.



In contrast, here's Very Sexy Seo.



First one to a Unicorn wins a prize.

This is just a sketch - very little logic used - just the RTweets being 'content analysed' under Bayesian tracking functions. (Finds the rarest word in the tweets, uses it for a tag to look up the image).

It's an easy (dirty) psychological trick to present an array of images as a form of narrative - the mind adapts itself to seeing a pattern regardless of how fuzzy it is.

The filtering from the characters file is informing the RTweets, and the loss of original context and thus original intended meaning does not carries over intact - leaving it open to interpretation - perfect for loosely coupling of images - as they too are devoid of their original intended meaning.

The new context provided by the Sibling demonstrates that it's the interaction with media objects that projects personality, not the definition of the character upfront.

As retrieval devices, perhaps they should evolve to be more 'refined' - which will involve a data storage. At this point, they remain more open to development oppotunities by insisting they exist in fleeting data storage (an archive in a twitter or FriendFeed stream).

Larger, permanent archives would give more data to process, giving a more refined feedback on how the character would/could/should develop. BUT, this would involve a sense of 'learning' and I'm not sure that is a wise path to follow. How, and by whom, should their mythology be maintained? Will their actions bear consequences? Will this provide destiny?

On a more important note, I don't think there is any 'family filter' on the slideshow. With respect to Flickr's TOS, I expect to see a little smut on the Siblings minds. Keep popping back to take a peek.


(Yes, this did turn up in a siblings feed.)


The Yahoo!Pipe used for the slide show is here.

Update - 10 minutes later...
I've added the Yahoo!Pipes output for VerySexySeo to its FriendFeed Account, with Friendfeed set to post any new images to Twitter. When the Sibling ReTweets, it forces Yahoo!Pipes to update, forcing FriendFeed to update (in real time) and post to Twitter. So, you should see links to Flickr images after a Retweet has been published, thus seeing the visual response from the sibling shortly after the Retweeting. The lag time should just be enough to make it feel human-ish.

Here's a FriendFeed Search for the images collected. They are posting to Twitter with URL pointing directly to Flickr.



The gray area is that users ReTweets are being 'transformed' without permission - but that's the semantic web for ya. Saying that, the ReTweet is published prior to the transformation, so attribution is paid up. I think we're on safe grounds with this.

Of course, this whole process can be applied to Youtube, Delicious, Technorati etc etc...2.0.

Which will be noisy, but helps blend the bots into the background (less spamesque) and harder to machine-detect against human managed account. But, they are on the edge of being global-micro-publishers.

voodoo

Number 1 Criss Angel fan, @magicangelgirl (Julie Smith), tweets a warming "Good morning to all" message, nestled within many shout outs to Criss Angel himself.



And fondly Retweeted by DGR M1 S1



And courteously Retweeted by DGR M1 S2.




Which is a bit of a surprise.

Not sure if that really is bots interacting with other bots, or not. But they're on the same wavelength it seems.

Voodoo.

Emotions run free - WFF API is back up

The WFF API kicked back in again this morning. All Siblings are getting their emotional resonances. Cron Jobs on Google App server are now running at 1 hour intervals for all Prototype Siblings, with a probability of '1 in 4' of them Tweeting - a variable to give the bots a slight randomness / human-ness / normalization to their eternal tweeting babble array based arias.



Now it's all about fine turning.

#MusicMonday

Some of rhe Test Siblings got caught up in the #MusicMonday meme on Twitter today.

Here's Jimmmo, tweeting not once but twice. First, he's going for The Passion by Lauren Hill ...





.. and I Liked You Better Before by Little Jackie.



Then a touch of Lithium by Evanescence at the end of some deep linking.





Very Sexy Seo come through with a selection of Michael Jackson classics.



>

Kranky Chloe votes for Dress Up in You by Belle and Sebastian. Which is a little spooky.





Even one of the early test bots gets in on the action.



Here's Love Drunk - Boys Like Girls.

Character defining applications

When all these twitter user grading applications turned up, the ones that define something of your character, the premise is 'you are what you tweet'. No sign of 'you are what you Retweet' yet. In the meanwhile, a Hot or Not application, Twitemperature, is entertaining.



A some what awkward moment.

Checksum Clouds

Wordle clouds are going to be handy in showing snapshots of a Siblings Retweeting in relation to their .yaml character definition file.

Here's ohlaylala's cloud from her top 20 words within her Tweets taken from the ohlaylala Twitter RSS feed. Compare that to her file. Bearing in mind that We Feel Fine is still down, and so the Retweets are based upon character's interests not emotional drivers.



Here's another Wordle cloud, but this time with the @replys removed and sorted via TweetStats.




Both are kind of meaningless at the moment, I think. Kind of interesting all the same.

A personality has begun.

The Science Of ReTweets

Here's a piece of classic Social Media analysis (presentation and video interview) from Dan Zarrella covering the spread-ability, readability and currency of the ReTweet. There'll be loads of this kind of thing on the DGR wiki in due course - just thought you'd like to see some context to the DGR project.

More comments about his report from Dan on his blog.

The Science Of ReTweets

View more presentations from Dan Zarrella.

For a more academic review of the Retweet, "Tweet, Tweet, Retweet: Conversational Aspects of Retweeting on Twitter" by Danah Boyd and friends, is a worthwhile read. It's currently only a draft, but it already contains gems such as:

Retweeting brings new people into a particular thread, inviting them to engage without directly addressing them. In this article, we argue that, as with link-based blogging [13], retweeting can be understood both as a form of information diffusion and as a means of participating in a diffuse conversation. Spreading tweets is not simply to get messages out to new audiences, but also to validate and engage with others.

And on the issue of shifting contexts:

Not all retweets are an accurate portrayal of the original message. When people edit content to retweet, they may alter the meaning of the original. Even when the content was not altered, taking a tweet out of context can give it a life of its own.

Here's the full draft text by Boyd and co. Enjoy.

Tweet Tweet Retweet by Danah Boyd

View more documents from David Bausola.

Track-a-Slice of the Demographics

Being able to slice through the siblings RTweets is Trackable.

FriendFeed supports live tracking; for example, here's 'love' and 'hate' being filtered out from the DGR Siblings aggregation.

Featured Bug

Just spotted VerySexySEO ReTweet it's own ReTweet. There's handling in the code to prevent this. Rob..?

FIXED. r30. 29/09/09

Just so typical of SEO strategies though, thus, making the bug a feature. Had to favorite the tweet for that alone.

Update. Ohlaylala gone and done it too. Notable, the one that was originally Retweeted had also been Retweeted by it's original author. This kind of thing really #GrindsMyGears.

Commemorative Video

I was sorting through the new assets and thought I'd process them through Picasa to generate a little commemorative video of the new faces. Music courteously from Morgantj.



In contrast, I found this on youtube, I must have done it months ago while sketching this project out.



The faces where done with Flash Face, and they all looked shifty. I kinda liked that at the time, but for a social audience, the new faces work, well, a bit better.

Music by: Well, I cant say. I added via Youtube's audio swap - and now I cant see the name of it. And it cuts out dreadfully at the end.

Friendfeed Aggregator

Here's all the public test DGRs in a real-time Friendfeed aggregation.



And when they share their feeds like this, you get to see the content, such as videos, pictures etc. Friendfeed makes a nice job of this. Here's ohlaylala Feed.

More TweetDeck sampling

A quick snapshot of the test DGR Siblings in TweetDeck.

Test results. Character matching

The above 3 sibling accounts (yet to be made public) are all using a copy of Kranky Chloes .yaml file. You can see how they all RTweeted the same message at the same time - due to the identical settings.

Note: TweetDeck is really good for keeping aggregated group views of the various Siblings.

DGR Model 3: Kranky Chloe

And another long term test subject Kranky Chloe.

http://twitter.com/krankychloe




Here's what makes Kranky Chloe do 'her' thing.

# All fields are compulsory

# The twitter user name and password
twitter_username: krankychloe
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of at least 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=0&agerange=40&city=london

# Emotional states
emotion_keywords: {
sad: [drunks, hippies, cats, shrubs],
great: [bus, umbrella, bycycles, ITV],
happy: [gin, vodka, red wine, white wine, banana, ITV]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
11: [tea, phone, gin],
13: [sarnie, gin, mustard],
18: [chips, spam, trolley, ITV],
22: [gin, sister, tesco]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [church, pub, neighbour, raining],
19: [mash, sausages, TV, BGT, ITV],
23: [sleep, tv, ITV]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords:
[Acton, grandson, gin, chips, ill, greyhounds, ITV]

DGR Model 3: Jimmmo

Long time test bot, Jimmo, just came out of private feed mode.

After introducing him to TwittGeek, he automatically hooks up with DGR Model 3 serial number 001. Go figure.




Here's Jimmmo's yaml.

# All fields are compulsory

# The twitter user name and password
twitter_username: jimmmo
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of at least 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=0&born=1972&city=london

# Emotional states
emotion_keywords: {
sad: [neighbours, TV, drugs, mum],
relieved: [mates, golf, cash],
happy: [beer, fighting, tattoo, gambling, football]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
06: [tea, cough],
07: [phone, boss, chase],
18: [dinner, beer, techno, facebook],
20: [girlfriend, mother, aunt, TV]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [practice, biking, bmx, wheels],
19: [dinner, beer, mates, pub, peas],
23: [sleep, tv, movie]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords:
[Hackney, Pringles, Ingles, greasy, pies, chips]

Handling Emotions Better

Rob has applied some more exception and error handling for the We Feel Fine API calls.

If the WFF api is down, like yesterday and today, the Sibling will now fall back gracefully onto key interests as specified in their .yaml files to complete the Twitter search and posting.

Cant beat live testing.

Above: ohlaylala spreading the good word.

We Feel (un)Fine

We Feel Fine API is off line for now, so all is paused. The test application has been slowed to every 5hours until it's back up.

oozzss gets frozen

This what happens if they tweet in public and have no followers or friends.

I'll drop the authorities a line about the DGRs.

Update.
http://help.twitter.com/requests/579891
Followed by an email saying the case was closed.
So I wrote asking why.

DGR Model 3: oozzss

Meet oozzss. oozzss is into techno.

# All fields are compulsory

# The twitter user name and password
twitter_username: oozzss
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of >= 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=f&born=1974&city=london

# Emotional states
emotion_keywords: {
happy: [techno]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
07: [techno],
11: [techno],
19: [techno],
23: [techno]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [techno],
19: [techno],
20: [techno],
21: [techno],
22: [techno],
23: [techno]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords: [techno]

DGR Model 3: VerySexySeo

Introducing VerySexySeo

An homage to Lora.

# All fields are compulsory

# The twitter user name and password
twitter_username: verysexseo
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of at least 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?returnfields=imageid%2Cfeeling&limit=50&gender=0&display=xml&city=london&agerange=20

# Emotional states
emotion_keywords: {
sad: [seo, sexyseo, verysexyseo],
great: [seo, sexyseo, verysexyseo],
happy: [seo, sexyseo, verysexyseo]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
11: [seo, sexyseo, verysexyseo, yahoo],
12: [seo, sexyseo, verysexyseo, google],
18: [seo, sexyseo, verysexyseo, ask],
21: [seo, sexyseo, verysexyseo, bing]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
10: [seo, sexyseo, verysexyseo, yahoo],
11: [twitter],
12: [seo, sexyseo, verysexyseo, google],
18: [seo, sexyseo, verysexyseo, ask],
21: [seo, sexyseo, verysexyseo, bing]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords: [seo, sexyseo, verysexyseo]

DGR Model 3: BMX Barry

Here's http://twitter.com/bmxbarry

And here is his particulars:-

# All fields are compulsory

# The twitter user name and password
twitter_username: bmxbarry
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of >= 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=1&born=1984&city=london

# Emotional states
emotion_keywords: {
happy: [beer, girls, bike, bmx, winning],
sad: [broken, crash, fallen, smashed, bent],
silly: [lads, chasing, running, tricks],
alone: [practice, cleaning, cooking, dreaming]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
07: [breakfast, eggs, tea, The Sun],
11: [practice, biking, bmx, wheels],
19: [dinner, beer, mates, pub, peas],
23: [sleep, tv, movie]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [practice, biking, bmx, wheels],
19: [dinner, beer, mates, pub, peas],
23: [sleep, tv, movie]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords: [crazy, brave, champion, dreamer, bmx]

DGR Model 3: ohlaylala

Here's one of the test Siblings.

http://twitter.com/ohlaylala



and can also be found here at Friendfeed.

http://friendfeed.com/ohlaylala

TwittGeek is now parsing the friend search, which will be carried over to Friendfeed - if successful.

Social too is also enabled to refollow followers.

The bot has been running for a week using a test.yaml character file, thus the grand number of tweets already.



The character file has now been customized as a bespoke persona.

Here it is.

# All fields are compulsory

# The twitter user name and password
twitter_username: ohlaylala
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of at least 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?returnfields=imageid%2Cfeeling&limit=50&gender=0&display=xml&city=london&agerange=20

# Emotional states
emotion_keywords: {
sad: [office, hippies, cats, shrubs],
great: [art, painting, colour],
happy: [creativity, optimism, painting, green, yellow, art]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
11: [drawing, coffee, radio, rock],
12: [salad, tomato, facebook],
18: [drawing, phone, facebook, painting],
21: [facebook, sister, biscuit]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
10: [fry up, gallery],
19: [coffee, talk, mates, party, pub],
23: [sleep, tv, radio]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords:
[creativity, optimism, painting, green, yellow, art]

More Face

Finally found a good face generator for the avatars.

http://www.morphases.com




Lot's of variety, nice editor plus you can upload your own face for remixing.
You can save the state of the design too - and share that around with friends.

All good.

Here's a quick test using the starburst effect, again.



Looking less sinister. Still in the uncanny valley though.

A little bluring around the edge of the head forces your focus onto the eyes / noise / mouth plane induces a emotional connection quickly - rather than allow the mind to take in the whole picture and doubt it's plausibility.

Mind, the avatar, when used on Twitter is only a thumbnail, so detail is lost to form.

At this size, the image becomes quite compelling.

Even at the smallest level, it works quite well.

Entertaining Augmention

One of the future milestones in the development plan is to not only being able to 'move' a Sibling around geographical markers, but show their position through visualisation techniques such as Augmented Reality layers.

Using services such as Dopplr and FireEagle via OAuth, Twitter accounts should be able to post locations where the DGR Sibling is. Where it will be will need to be generatively plotted, using the routine arrays in the DGR character .yaml file.

At present Locations data direct to Twitter accounts is a bit to open (you can state anything - London, around the corner, over the bridge, edge of nowhere) which isn't too useful. More detailed, controlled approach is coming.



View A DGR Stroll in a larger map

Using Google maps as a tracks builder might be a good way to proceed for regular routine activity, or perhaps, as macros to execute certain routines, such as going to the pub and staggering home again. Above is a Google Map with 2 'Strolls' plotted. This should be enough information to set a Sibling off on a routine - complete with timings.

FireEagle based applications such as EagleFeed and FiredUp seem like good integration services that are firmly OAuth based.

As a broader, exploratory, exercise, it would be nice to let the Sibling free to roam the planet - so that it explores, RTweets from within it's geographical contexts, "gets lost", finds it's way back home, or goes and hangs around other people that it find a connection too.

Anyway - all to come in due course, just thought I'd mention where we think this going.

Here's a video clip from Layar, demonstrating how 3D models can be plotted along geographical data arrays - such as a flight path. That should give you enough to realise that the Siblings should be able to wander around the streets and be visible through your phone's camera lens, enabling you to watch their live action unfold.

Using FriendFeed to see the content

Once logged into your Sibling's Twitter account, OAuth makes it easy to extend the you character into other services.

Friendfeed is thus a one click subscription, plus it finds who is following your DGR on Twitter and who is already Friendfeed and adds them to your feed. Which is nice.

Friendfeed then shows you their messaging, PLUS content from all their other streams. You end up with a mix of content relating to the Siblings character description in their .yaml file.

It's an odd mix. Here's a sampler of Youtube films from DGR Model_3_02 Friendfeed from this afternoon.

Avatar Experiments

Random Flash Face with Star bursts.



Monoface with Starburst.



A Flashface of Monoface



Flashface of Monoface with Starbursts. And drop shadow.



Inspired by @ribrob

Facial representation

Creating the Sibling's avatar needs to lean towards fun.

Using Flash Face give them a sinister edge - a bit sinister.

Trying to find a process that 'generates' fun faces is seemingly hard to track down.


They have to get over the uncanny valley and be easy to (mass) produce - automated would be perfect.





A few rules:

It has to be a face not an graphic icon.
- they don't seem real.

Manga'isation
- as per graphic icon

Has to be unique.
- per personality

Has to be easy to make
- though spending hours working on your avatar is not out of the question.

Using CC licences photos of faces? Not Sure.

Some Links

Autoface generator using photos. Good but doesn't scale to 1000s well.
Monoface

Cute Manga figures. Too Graphical.
RasterBoy

Criminals...mostly. Black and white photo fit. Use with care,
FlashFace

Starburst with vectors graphics.
Tutorial and samples

iPhone 'shake' generator.
Nice. Haven't tried it though.

faceyourmanga
As it says.

Find Friends for the Siblings

As an aid to naturalise the DGR Siblings, they need people to follow. This should lead to some work on processing tweets from who the Sibling follows and then tweeting something appropiate. May this be some generative from followed tweets or else is yet to be decided.

TweetGeek is being trialled with DGR Model 3 02 as a following tool for the Sibling. It uses OAuth, which is handy.

It adds 200 people based upon keywords.

For this trial I've used the base keywords from the DGR M3V2 .yaml file.

character_keywords: [crazy, brave, champion, dreamer, bmx]

From the initial collection of friending of 200 accounts, around 5% started following DGRv3m2 - unsurprisingly, they we're mostly spam or social media 'gurus'.

It would be good to find a similar befriending application based upon the RTing that the Sibling is doing. Ergo, make content from spam. Spam Radio is a fine example of this.

But back to the account in hand.

Following: 194
Tweets: 20
Followers: 10

Seems quite nicely rounded.

DGR Model 3: 002

This is DGR Model3 Serial number 002.

The bot posts from this account:

http://twitter.com/DGR_m3_02

DGR m3 s2 is a public test bot.

The DGR parameters are:

# All fields are compulsory

# The twitter user name and password
twitter_username: DGR m3_02
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of >= 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=1&born=1984&city=london

# Emotional states
emotion_keywords: {
happy: [beer, girls, bike, winning],
sad: [broken, crash, fallen, smashed, bent],
silly: [lads, chasing, running, tricks],
alone: [practice, cleaning, cooking, dreaming]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
07: [breakfast, eggs, tea, The Sun],
11: [practice, biking, bmx, wheels],
19: [dinner, beer, mates, pub, peas],
23: [sleep, tv, movie]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [practice, biking, bmx, wheels],
19: [dinner, beer, mates, pub, peas],
23: [sleep, tv, movie]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords: [crazy, brave, champion, dreamer, bmx]

DGR Model 3: 001

This is DGR Model3 Serial number 001.

The bot posts from this account:

http://twitter.com/DGR_m3_01

DGR m3 s1 is a public test bot.

The DGR parameters are:

# All fields are compulsory

# The twitter user name and password
twitter_username: DGR_m3_01
twitter_password:

# Doesn't change. make sure it has imageid and feeling an a limit of at least 20
wff_search: http://api.wefeelfine.org:8080/ShowFeelings?display=xml&returnfields=imageid,feeling&limit=50&gender=1&born=1972&city=london

# Emotional states
emotion_keywords: {
sad: [drunks, hippies, cats, shrubs],
great: [bus, umbrella, bycycles, TV],
happy: [gin, vodka, red wine, white wine, bananas, TV]
}

# Times of day (after given times), 24hour clock, 00..23
time_keywords: {
11: [tea, phone, gin],
13: [sarnie, gin, mustard],
18: [chips, sausgaes, TV],
22: [gin, sister, aeroplanes]
}

# Times of day for Saturday & Sunday (after given times), 24hour clock, 00..23
weekend_time_keywords: {
09: [church, pub, neighbour, raining],
19: [mash, sausages],
23: [sleep, TV]
}

# Words to fall back on if nothing else matches, or to add randomly
character_keywords:
[London, grandson, gin, chips, wobbly, greyhounds, backgammon]