Hit interestingness scores (phase 2)

Concerns? Let us know by posting here.

Moderators: Fons, avij, Phaseolus, dserrano5

User avatar
lmviterbo
Euro-Master
Euro-Master
Posts: 4151
Joined: Thu Aug 21, 2003 5:23 pm
Location: Lisboa, Portugal
Contact:

Re: Hit interestingness scores (phase 1)

Postby lmviterbo » Mon Feb 25, 2013 7:58 pm

A hit on a dead straight line is quite interesting because it is quite unusual. That may be scored on deviation from 0º/180º or 90º/270º.

In fact, I like the way R/J has put it and I believe this could be an important key for every criterion of interestingness: deviation from average.

Also, it would be superneat if it's possible to include historical meaningfulness to hits without changing interestingness rating over time. Here's an example:

avij wrote:How many hits the users have in common? If it's their first hit, it's more interesting than if it was their 500th.

Let's suppose I have my first hit with avij. This would be rated 25, for instance. When we have our second hit, it would be rated, say, 15. Now the first hit shouldn't change its rating because of a second hit. In fact, I guess this is exactly what you are thinking but I wanted to clarify it.

As to bundle hits, I can't really say that in a bundle of 30 hits the first that was entered is more meaningful than the last one, just a split second later. Bundle hits might need a special treatment so that they all keep, for example, an average between the rating the first one would receive and the rating for the last one.

[edit/avij: replied to the other topic, as I think the bundle hits problem may need broader discussion]

User avatar
dserrano5
Euro-Master
Euro-Master
Posts: 3773
Joined: Sun Jan 31, 2010 11:59 am
Location: Madrid, ES

Re: Hit interestingness scores (phase 1)

Postby dserrano5 » Mon Feb 25, 2013 11:28 pm

lmviterbo wrote:In fact, I like the way R/J has put it and I believe this could be an important key for every criterion of interestingness: deviation from average.


Beware, that's deviation from average at the moment the hit happens. Future hits will alter the average (perhaps in a significant way) but that won't change the scoring of past hits (otherwise, this would mean that the scores for each and all past hits would have to be recalculated every time a new hit happened).

User avatar
avij
Forum Moderator
Forum Moderator
Posts: 5614
Joined: Mon May 27, 2002 10:45 pm
Location: Helsinki Finland
Contact:

Re: Hit interestingness scores (phase 1)

Postby avij » Sat Mar 09, 2013 7:31 pm

When you're travelling, the hits you get while on the road are interesting, right? I believe this case isn't yet fully covered by our current list. Perhaps we should also consider how many notes the users in question have entered in each city/region/country to determine if they're travelling outside their usual habitat. Example: A Finnish tracker visits Munich and gets a hit with only a relatively few entered notes in that city.
Money makes the world go round. We track how the money goes round the world.
EBT Tech WG leader. Do not PM me if your question is not related to Tech WG or the association.

User avatar
lmviterbo
Euro-Master
Euro-Master
Posts: 4151
Joined: Thu Aug 21, 2003 5:23 pm
Location: Lisboa, Portugal
Contact:

Re: Hit interestingness scores (phase 1)

Postby lmviterbo » Sat Mar 09, 2013 8:12 pm

On another topic now obsolete, Craft has told a story of a very interesting hit:
Craft wrote:[…] greetings from München, I just got a hit with a :note-10: that I got as change from a ticket to the Olympic tower today at noon. It cost €5,50 and I paid with a :note-100:, getting back 2x :note-10: , 1x :note-20: and 1x :note-50: . My first hit with a note that I have received on my trips abroad. Really surprised and happy :D. For me, this really is an interesting hit! Thinking about the probability of getting a hit when I handle altogether maybe less than 20 notes during my stay here.

The whole story made me think of one additional criterion for interestingness (I don't recall this having been mentioned before like this):
  • Location of the ebtist at the moment the note was registered — as always, taking into account the "deviation from average" general criterion. A hit with a note at home is less interesting than a hit on a travel. Even more interesting if the region is different, even more if the country is different.

Craft wrote:[…]And even more incredibly: I only now noticed that the :note-100: that I paid with was itself a hit. I took the note from a bank in Finland along with a few others before this trip and entered them all quite hastily in Finland, but I didn't even notice then that there was a hit.

So, in a perfectly normal payment situation today, I unknowingly changed a hit note to a bunch of smaller notes, one of which was a hit note. I noticed this all only now in the evening. I could say that this is maybe - or actually, definitely - the most interesting hit day for me so far in my whole EBT history. Coooolll 8) .I bet the new interestingness score cannot take this kind of interestingness into account. :lol: . It would require data (about notes involved in each individual payment/change transaction) that is not recorded into the database.

I have always wanted to know and register where I spend my notes, because that would be the way to really track the note in all its controllable path. Unfortunately, no one has devised a practical way to do it, yet. I am not a heavy user, so I can and do add some additional comment in cases like this one. Whenever I take a hit note elsewhere, I edit the note's comment to inform all others where I've spent it. I also like to do it on notes I register abroad and bring back to Portugal.

This kind of interestingness derived from personal experience (already in comments when the hit is produced, or revealed later) should be computed somehow in the score, but I believe this was already mentioned. I mean: the ebtists involved in a hit should themselves be allowed to classify a hit according to its interestingness for them.

One way of doing this could be adding a new entry field just after you receive the system message telling you've got a hit. So messages m_4_billentered_with_hit and m_4_multi_entered_ok_with_hit would be rewritten to warn you not only of the hit but also that you can classify and comment the hit. The same applies to the hit notification email message (M_HIT_EMAIL_HELLO). The classification could be a simple 1-to-5 stars thing, and the comment would be added to your own note's comment.

By the way, I have just changed the translations for these constants, because in Portuguese the word hit wasn't mentioned here. Translators and Development Team, you may check my other post about it here.

User avatar
dserrano5
Euro-Master
Euro-Master
Posts: 3773
Joined: Sun Jan 31, 2010 11:59 am
Location: Madrid, ES

Re: Hit interestingness scores (phase 1)

Postby dserrano5 » Sat Mar 09, 2013 8:33 pm

avij wrote:When you're travelling, the hits you get while on the road are interesting, right?


Not sure to agree. If I travel to Finland and enter 1000 notes while being there, I can reasonably expect to get some hits. But this is in fact less interesting than a) not getting any hits at all with 1000 notes :D and b) hitting Finland without travelling there.


avij wrote:A Finnish tracker visits Munich and gets a hit with only a relatively few entered notes in that city.


But this is interesting, to me at least, because Craft entered only a few notes in Munich (way less than the city's hit ratio, as I understood it), not because he entered them while away from home.

User avatar
Craft
Euro-Master
Euro-Master
Posts: 1182
Joined: Fri Jun 07, 2002 4:36 pm
Location: Vantaa, Finland

Re: Hit interestingness scores (phase 1)

Postby Craft » Sat Mar 09, 2013 9:10 pm

Just for the record, regarding hit ratios:
In three days I have entered seven notes so far here in Munich. Tomorrow is my last day here, maybe some more notes but not many.

dserrano5 wrote:
avij wrote:When you're travelling, the hits you get while on the road are interesting, right?

Not sure to agree. If I travel to Finland and enter 1000 notes while being there, I can reasonably expect to get some hits.

Hmm, provided that the number of notes entered while travelling is taken into account in the formula (meaning: the more notes entered, the less interesting a hit becomes), then I'd say hits on the road are more interesting than hits back where you usually enter your notes. When I was planning this trip and thinking about entering notes here, I didn't even play with the thought of getting a hit here, knowing in advance that I will enter much fewer notes than my hit ratio suggests for making a hit probable. I didn't come here for trying to get a hit - actually I was much more interested in getting two or three new dots to my map.

I think that most users would agree on the higher interestingness of hits somewhere else than their usual habitat. The other way round, I'd expect that if the travel factor were left out of the interestingness formula, users would question the reason for this decision. I regularly enter notes where I live and work, but coming to Munich for just a few days and getting a hit here is something special, not least because Munich will now surely stay in my memory - in addition to its other virtues - as a place of a hit abroad. I suppose that many would feel likewise in similar situations.

OTOH, the interestingness of this type of hits doesn't need to be exaggerated. I'd say travel hits should be graded as more interesting than home hits, but to some extent only. And there might be other factors effective at the same time, adding to (or subtracting from) the travel hit's interestingness, the number of entered notes being one but not the only one of such factors.
Last edited by Craft on Sat Mar 09, 2013 10:23 pm, edited 1 time in total.

User avatar
Craft
Euro-Master
Euro-Master
Posts: 1182
Joined: Fri Jun 07, 2002 4:36 pm
Location: Vantaa, Finland

Re: Hit interestingness scores (phase 1)

Postby Craft » Sat Mar 09, 2013 11:18 pm

BTW, which point of view would be easier for this development:
- What are the factors that make hits more interesting?
- What are the factors that make hits less interesting?

More precisely: should we consider a formula that makes both additions and subtractions to a hit's interestingness? I'm a bit worried about if we only list factors that make hits more interesting, but ignore factors that make them less interesting. A single hit might possess both kinds of factors so that its score lands to an average level despite some high interestingness factors, whereas another hit might have the same high interestingness factors but lack the uninterestingness (this word might not exist in real English :lol: ) factors and thus properly get a higher score.

Of course, many of the listed factors may act as both interestingness and uninterestingness factors, depending on their value. But there might exist factors that are only one or the other.

User avatar
avij
Forum Moderator
Forum Moderator
Posts: 5614
Joined: Mon May 27, 2002 10:45 pm
Location: Helsinki Finland
Contact:

Re: Hit interestingness scores (phase 1)

Postby avij » Sat Mar 09, 2013 11:42 pm

I'd generally concentrate more on promoting than demoting. But if you can come up with some ideas for uninterestingness that are not simply the opposites of the qualities of an interesting hit (those that we've already mentioned in this topic), I'd be interested (pun unintended) in hearing about those.

Craft wrote:the interestingness of this type of hits doesn't need to be exaggerated

Certainly true. The idea is that we can assign a weight for all the individual metrics. This allows us to include even some generally unimportant factors in the calculation, which may or may not nudge the "star rating" just above/below the threshold for the next/previous level. You'll all get to have your say about these weights later on, I have a nice interactive system in mind where you can test out the various weight parameters yourself. If you think something shouldn't be included in the score calculation at all, you can assign it a weight of zero. But this is for the later phases, we're not there yet.
Money makes the world go round. We track how the money goes round the world.
EBT Tech WG leader. Do not PM me if your question is not related to Tech WG or the association.

User avatar
Craft
Euro-Master
Euro-Master
Posts: 1182
Joined: Fri Jun 07, 2002 4:36 pm
Location: Vantaa, Finland

Re: Hit interestingness scores (phase 1)

Postby Craft » Sun Mar 10, 2013 9:35 am

avij wrote:But if you can come up with some ideas for uninterestingness that are not simply the opposites of the qualities of an interesting hit

I was thinking if there are factors whose one end is quite negative regarding interestingness, but there isn't any extremely positive other end, i.e. the other end is just "normal" or "commonplace". I'll try to find an example later.

But this might of course always be a matter of definition: you could say that a factor's highest (= "normal" at best) case is 0 and uninteresting case is -10, but you can also define that the highest case is +10 and the uninteresting is 0 (like the Kelvin and Celsius scales are essentially similar, only the 0 point is set differently). I know this is not the metrics phase yet. Just trying to analyze if we are about to ignore some factor(s) that could actually partly or totally cancel a hit's interestingness which it first seems to get based on the other factors.

For instance, a hit might seem very interesting by many factors, but it is then revealed to have happened due to a factor that makes it dull after all. A case I can think of: Let' assume that I live in a neighbourhood where there is a fellow ebter that I know of (even though we have never met in real life or even virtually) and we often get hits with each other just by chance. These hits are quite uninteresting. Then I travel to, say, Munich, get a lot of notes there (I don't, but let's assume) and get a few hits there. I then keep at least some of the hit notes, not necessarily intentionally, come back to Finland, and use the notes there in my neighbourhood for normal payments and such. Then I again get hits with this fellow ebter with these notes. I didn't bring them to get hits with this fellow, I just cannot avoid getting the dull hits with him unless I make special arrangements to get rid of all the hit notes before coming back to Finland. Now these hits might seem quite interesting because they are triple which by definition is going to be more interesting than a simple hit, and because they are international. But actually they are not that interesting. The fact that I transported hit notes from Germany to Finland could be regarded as an uninterestingness factor (giving minus points). The other end - i.e. I leave the notes to Germany to avoid dull hits in Finland - does not actually add to the interestingness, if these notes get hits somewhere later on. Only if my neighbourhood's fellow ebter gets a hit with one of these notes somewhere in the future even if I definitely didn't transport the notes to Finland, that would certainly be extremely interesting!

So, transporting hit notes unintentionally and getting further hits with them could be regarded as an uninterestingness factor, whereas not transporting them doesn't seem much of an interestingness factor as such but just the normal case. (Intentional transportation and feeding them to a fellow ebter is another matter that falls to the abuse category.)

But I think these reflections go off the topic :? so please feel free to delete these posts (I don't have the rights to do it myself).

savesigita
Euro-Regular
Euro-Regular
Posts: 165
Joined: Sun Sep 30, 2012 2:38 pm
Location: Italia + Latvija +wherever we get the chance to travel to!

Re: Hit interestingness scores (phase 1)

Postby savesigita » Sat Mar 16, 2013 10:18 am

avij wrote:First phase: Determining all the qualities of an interesting (or uninteresting) hit

So, what makes a hit interesting? I'd say that it's unusual. That's a blazingly obvious statement, but it actually helps with the next definitions. There can be some overlap with the qualities, for example it's likely that a hit that has travelled thousands of kilometres is also an international hit, but that's okay. We can still include both in the calculations.

Comments
Comments can't unfortunately be used for calculating the score automatically

Is there something else that you look for in an interesting hit?


but the point about the comments it's really bad!
Where or how you find a banknote even if it makes a hit with few km, or few days can be very unusual!
for example, I've found 4 banknotes on the floor so far! they weren't hit, but if one was, let's say, Firenze-Firenze days 50 km 0, would you think as usual a hit made with a flying banknote found on the floor?! and there are many others CASUALITIES like this that make a hit very unusual, would't be unfair to put them at the same level of other grey, boring, standard hits?

mizar
Euro-Regular in Training
Euro-Regular in Training
Posts: 128
Joined: Tue Jun 05, 2012 5:18 pm
Location: Slovenia

Re: Hit interestingness scores (phase 1)

Postby mizar » Sat Mar 16, 2013 10:25 am

avij wrote:First phase: Determining all the qualities of an interesting (or uninteresting) hit

Comments
Comments can't unfortunately be used for calculating the score automatically
<< this is the key word I believe. Of course notes found in strange circumstances are very interesting, and they can and should be featured e.g. in dedicated forum threads, but it's pretty much impossible to put this into an algorithm. Besides determining the "value" of each situation, someone would have to manually check all the comments (in different languages) and find the interesting ones ...

savesigita
Euro-Regular
Euro-Regular
Posts: 165
Joined: Sun Sep 30, 2012 2:38 pm
Location: Italia + Latvija +wherever we get the chance to travel to!

Re: Hit interestingness scores (phase 1)

Postby savesigita » Sat Mar 16, 2013 10:53 am

mizar wrote:Of course notes found in strange circumstances are very interesting, and they can and should be featured e.g. in dedicated forum threads, but it's pretty much impossible to put this into an algorithm.


Then comments become meaningless! Original circumstances will be banned, or in your words "interesting for dedicated forums"! but still when you look at the stars based on interestingness and unusualness your special hit will be neither interesting nor unusual, fair?

mizar wrote:[ Besides determining the "value" of each situation, someone would have to manually check all the comments (in different languages) and find the interesting ones ...


we are talking about a big revolution for EBT, i don't think people who want to make it are thinking about how long it will take!

savesigita
Euro-Regular
Euro-Regular
Posts: 165
Joined: Sun Sep 30, 2012 2:38 pm
Location: Italia + Latvija +wherever we get the chance to travel to!

Re: Hit interestingness scores (phase 1)

Postby savesigita » Sat Mar 16, 2013 11:26 am

avij wrote:Comments
Comments can't unfortunately be used for calculating the score automatically


Interesting is not just the banknote itself but the story that carries with itself! and comments are a necessary part of this story!

But there is a solution:
At first the algorythm of all the parameters already decided can be normally calculated, but after it can be offered a possibility, to the owner of a hit, to ask support, tell the reasons (connected to the special casualities i wrote above) why he thinks that the hit deserves more stars and then moderators can decide if upgrade or not the interestingness of the hit!
Of course it will be a bit subjective decision of the moderator, but when they moderate hits they often make decision with a level of subjectivity, so there wouldn't be anything wrong!

User avatar
dserrano5
Euro-Master
Euro-Master
Posts: 3773
Joined: Sun Jan 31, 2010 11:59 am
Location: Madrid, ES

Re: Hit interestingness scores (phase 1)

Postby dserrano5 » Sat Mar 16, 2013 12:13 pm

savesigita wrote:At first the algorythm of all the parameters already decided can be normally calculated, but after it can be offered a possibility, to the owner of a hit, to ask support, tell the reasons (connected to the special casualities i wrote above) why he thinks that the hit deserves more stars


If users are given the chance of asking their hits to be given more score, then they will overwhelm the support team with pointless requests. IMO users should only have an option to downgrade the score.

Re taking the comments into account, I'm against that. What we are developing now is already a substantial improvement over the current system and I think comments won't (usually) further improve the results. Don't forget that comments are already being ignored now and we've been EBTing for 10 years without that causing problems. Plus, it's a remarkably difficult challenge to tackle and the effort/reward ratio is certainly low.

User avatar
avij
Forum Moderator
Forum Moderator
Posts: 5614
Joined: Mon May 27, 2002 10:45 pm
Location: Helsinki Finland
Contact:

Re: Hit interestingness scores (phase 1)

Postby avij » Sat Mar 16, 2013 12:27 pm

Yes, "automatically" was indeed the keyword.

This is what I wrote in my first message to this topic:
avij wrote:There may also be other reasons for increasing or decreasing the score slightly based on the user's own knowledge of the hit.

I think this covers the information that is recorded in the comments. One of the components for the score calculation can be user's own rating of the hit (either positive or negative). It will be just one metric among the many others, with an appropriately assigned weight (to be determined at the later stages). We'll also need to limit the usage of this capability, ie. it shouldn't be possible for a user to declare that ALL his/her hits should deserve the interestingness boost. The details of this can be discussed in the later stages.

savesigita wrote:to ask support, tell the reasons (connected to the special casualities i wrote above) why he thinks that the hit deserves more stars and then moderators can decide if upgrade or not the interestingness of the hit!

This is exactly what I've been trying to avoid. We'd end up again in a situation where the upgrades/downgrades would be arbitrary.
Money makes the world go round. We track how the money goes round the world.
EBT Tech WG leader. Do not PM me if your question is not related to Tech WG or the association.


Return to “Feedback and Development”