Context-Aware Computing - Science topic

Share price forecast comes into mind.

If you have a time series of values,

use a gliding window over some past

data as training set for a feed-forward

neural network. Say a window with n values.

Take k values of the window as input

vector, n-k values as output vector.

In your case some compression should

take place in order to avoid big

vectors. For example an a user

activity over 30 minutes should not

emit a signal every second. You

could model it by boolean values

meaning "for at least 10 seconds",

"for at least a minute",

"for at least five minutes"...

Later you will have to extract rules

from the neural net. That's a

different task, but feasible.

Regards,

Joachim

See publications below if you are still interested in the topic.

lala protege ko achay tareekay say dayk lo or google ostaz say apnay problems baghair kuch chpayay pocha karo.

Not so easy to answer. However, one idea - if I understood you right - would be:

based on your GPS environment, filter wikipedia using the GPS and the category mechanism in wikipedia to select whatever you want in your environment.

 Ed, I think that you have an interesting question here.  There are numerous indoor activities that could benefit from more fine-grained position awareness (trade shows, conferences, schools, etc.) and I think that it is just a matter of time before we see more and more iBeacons around.  The question that I have about this is the logistics of maintaining the infrastructure.  Any kind of sensor needs power and even with the best battery life that we currently have there is a fairly short time limit for the utility of the beacon.  If we start "throwing" lots of these into the environment what will the near future look like when the batteries start to fail?  I see some work on the use of these sensors and beacons but have not seen much on the long term prospects for the technology - IMHO this will be a big issue for Internet of Things.

You should read this survey "Contextual music information retrieval and recommendation: State of the art and challenges" by Kaminskas and Ricci. Other papers not directly related with your question but useful may be: 

Hope it may be useful for you. 

Cristhian. 

That is a good point Krzysztof, one cannot 'impose' a desired mood. Some studies suggest (i don't have any citing at hand but i will find it) that when you are depressed you don't want to listen some happy music, most probably you will prefer sad music. I don't know if that will help on your mood... In the case of gaming we need to look to other aspects; if a game don't activate you then it must be in a boring stage, as a programer you need to foreseen that case and prepare some action, whatever that means in the context of the game. So addressing your question i would say that It Depends ! if you want a player to engage more in the game you need to activate hime. If you want someone to let go a depressive state (in ehealth apps) you need first to get in the mood and then try to cheer up that person. But if you are developing an app for a business man, if you detect high stress rates, better not to ring that high tone ring alarm, right? I don't think there is a yes/no, black/white kind of answer to your question, it depends on context assessment. Do you understand my point and do you agree?

The question is too fuzzy for accurate answers, but here's some more ideas about context. Many of these current answers are very limited and narrow - context is a huge term that is used in numerous alternative ways. You need to be more precise what you want to achieve with context and modelling. What you mean by context and modelling. Etc.

Research about provenance is all about documenting and tracking context. This article reviews are huge amount of multi-disciplinary provenance (=contexts) research. The topic is huge.

One clearly a methodological paper about context modelling is this: CEIP Maps: Context-embedded Information Product Map

There is a lot of research about context in HCI research. The whole discipline is based on the concept of context. For example HCI ISO standard defines Context of Use as a combination of users, tasks, tools and environments. You can model these issues in unlimited level of detail and complexity. There exists a lot of managerial and technical tools for modelling contexts in user centred development processes. Software systems for modelling context as well as systems to monitor contexts. Etc.

Hi Ines, It is an interesting question. It would be even better if we had some examples or diagrams here to illustrate the issue. I teach (and do) systems modeling and many of of my students also do context diagrams, aside from data models, process models and object models. My general answer would be yes, it is sometimes possible to have the same concept in different models. If something is in the business context model but is  also a major entity that we want to record detailed data on, yes, it would make sense to have it in the data or database model as well. Location related data, for example, can be quite detailed. Best wishes and good luck with your work. 

You can also have a look at Alcaraz and Lopez, who discuss the differentiation between situation awareness and context awareness:

Alcaraz, C., Lopez, J. (2013) Wide-area situation awareness for critical infrastruture protection. Computer 46(4). pp 30-37

Although latest mobile devices are quite good with resources but the limitation of resources to run applications is still a problem. The applications should be designed with following perspectives in mind.

1- Make sure that any resource like memory or process that is not being used anymore should be release. It can turn up to be bottleneck on mobile device soon compared to normal computing devices like computers.

2- Quality testing is very important. Any process or feature that uses memory should be watched to see if it actually has a requirement for increased memory as it runs. It should be tested with this application running alone or with other apps running in background and see if there is any difference in frequency of memory increasing. Decision has to be made if it actually needs to store all that information or it can keep a snapshot and load details when required or demanded.

3- Battery consumption is a major problem in apps. They can be wonderful but useless if they eat battery. Usually it is done by keeping those threads/processes alive when it is not really required or the interval between them is unnecessarily shorter making it heavy for battery. Also it can be seen easily if the application is tested at least for 24 hours - that should tell the pattern of battery consumption. The functions should be decomposed to see what exactly is causing that and then algorithm should be analysed if it can be optimised.

Hope that helps.

I am not talking about all types of context information, only talking about 'relationship context'. I am wondering different types of relationship context (e.g., interpersonal relationship)? What will be the proper sources to collect this information? What will be the other low-level information to infer relevant relationship context information?

A sample motivating scenario from the healthcare domain for the patients medical records management can be found in our paper. In our paper, we have briefly discussed a context-aware scenario, that illustrates the need for the incorporation of context information in the access control processes. Also, you can have a look on the Prototype section.

It is also my impression that context-oriented programming encompasses feature-oriented programming in the sense that you could consider (certain) contexts as features, and on the condition that you have a context-oriented language which provides support for declaring (and verifying at run-time) feature-like dependency relations between contexts (as is for example the case in the Subjective-C language). But what about my other questions :

Can feature-oriented programming learn from context-oriented programming to enable for more dynamicity?

Can context-oriented programming take inspiration from feature-oriented programming and software product lines to offer more control and guarantees about the program's dynamic behavior?

Should/will they eventually merge into a single paradigm that takes the best of both worlds?

The users motivation is informed by context, and the context is fedback from the motivation in such a way as to inform the context of the motivation. But motivation is tricky, it is not as simple as the model we call Will would seem to suggest. For one thing Libetts work seems to suggest that the motivation often comes before awareness is practical. I am a proponent of the Weak Attention model, where attention does not bow to will, so much as will is an illusion of unconscious processes, that feedback through awareness the impression of causality linked to the report, rather than to the unconscious response. Thus attention is not just the spotlight of concentration, but also the less obvious but no less important control that directs memory between processing stages within the mind, weakly linked to will.

Yes, the connection is quite clear - the benefit of having digital memory for context trends crawling is hard to underestimate.