Nivedit Majumdar Nivedit Majumdar

The Many Spectra of Context Awareness

Technology is improving by leaps and bounds, and one key trait is shared by most developments in the various verticals of the tech landscape: devices and software are now becoming enabled to intelligently sense the user’s requirements. Be it in the form of smart voice assistants, or even smart home devices under IoT, there is a paradigm shift in the way machines interact with users.

A driving force behind this development is the involvement of context. Context is the fundamental building block of concepts such as machine learning and artificial intelligence, and we’re just at the tip of the iceberg here. Big data analytics, data mining, smarter application interaction and even lifelogging – all these sectors are seeing tremendous potential of Context Awareness.

And that is the crux of my article here. I discuss all the various aspects of Context Awareness and the fields it is making its presence felt in.


There’s always a beautiful symbiosis existing when it comes to context and data. While data itself is available in plenty, there needs to be recording instruments to take in all the data. And also, the data that is recorded needs to be relevant, and not merely blocks of intangible, random information.

This is where context steps in. It enables a recording instrument (we call these instruments sensors!) to effectively record the parameter that is beneficial and relevant to the user’s actions at that particular moment. Through context, the data that is captured can have a true purpose at a later stage. And isn’t this the underlying focus of lifelogging and the Quantified Self movement.

Earlier, content used to be king. But now, this adage can be modified to state that context is king. Diverse data can be aggregated together to ensure meaningful information can be derived, which later on paves the path for improved insights.


Data mining is an interesting branch of data analytics, which revolves around the concept of extracting useful snippets of data from larger pools of information. While this branch itself has been efficient to a certain point, it does bring about the question, how can it be improved further?

I was reading an interesting paper by Pravin Vijarkar et al on Context Awareness in Data Mining, and one thing is for sure: Context is the factor which can exponentially improve the efficiency and efficacy of data mining. Through context, undesired factors or results can be seamlessly filtered out to make sure that only the relevant data is left behind.

The paper discusses multiple types of context, which primarily include:


Which depends on the user’s level of expertise and knowledge. This in turn is governed by the past readings or experience of the testing individual.


This is the branch of context pertaining to the application level, which may or may not incorporate the wireless features. This in turn is of two types: Joint Conference Context, which enables data from multiple streams to cross paths, and Time Context, which also takes into account the time of recording the data, and therefore the presence (or absence thereof) of environmental and external factors.


This involves the clustered datasets that are recorded, and then re-analysed upon. It involves verticals such as Domain Context and Location Context, which brings the element of correlation into the equation – that is correlating data according to location based readings to come to a valid conclusion.

Now, the million dollar question is this: How can data mining help in gathering relevant information? And also, how can context aid data mining?

The underlying concept is quite simple and elegant, really. The presence of context awareness in data mining related applications will ensure that patterns in data are detected. The useful and interesting trendlines are isolated from the aberrant numbers, and this data is compiled together to give rise to information through application level and data level Context Awareness.


Converting this information into insight? That is primarily where Human level Context Awareness steps in, but nowadays Application and Location based Context Awareness are stepping up their game. A notable example would be Tesla’s self driving cars, which will automatically learn your driving routes and accordingly adjust the driving dynamics.

Moving ahead from data mining based applications, let’s take a look at the core Application level now.


Okay, so this is where machine learning and artificial intelligence come into play. Applications study the user’s habits and will accordingly make suggestions that would better suit the overall user experience. And all this, through an intricate system of context awareness.

We are subjected to simple forms of context awareness in applications even without being aware of it ourselves. Things like YouTube suggesting videos based on your viewing history, eCommerce sites suggesting products based on your product browsing history, auto-suggest on your phone predicting which word you would want to type next: all these things are based on one holistic concept: Context Awareness.

Context Awareness is definitely going to be big. It’s already making its presence felt in smart assistants such as Siri, Cortana and Google Now. In fact, Context Awareness is one of the important sciences behind the development of our application Launchify – which studies your app usage and automatically places frequently used applications in a convenient position, based on suggestions from your app usage, places and driving.

Launchify app

But more than applications and software, let’s talk about things closer to the Quantified Self movement.


Now this is where things become really promising. I’ve always maintained that a correlation of all the data that is recorded is key to gaining a better perspective of how the body functions, and also making sense of all the numbers that are being recorded. And in this regard, there is nothing more crucial than the aspect of context awareness within lifelogging.

Context Awareness in Lifelogging as of now comes in the form of various applications within a phone gathering data through a multitude of sensors, which in turn communicate with each other. An aggregating service taps into these data streams and contextually filters out trendlines from the useless data, and helps the user make sense of statistics.

The scope is huge, the potential of sensors and algorithms are ever growing. I personally predict that lifelogging and the Quantified Self movement are two major spaces which will see a lot of growth (and in turn contribute to the growth of) because of Contextual Awareness.


Long story short, Context is king. To enable applications and hardware to become truly smart, and to user in eras of semi-sentient technology, artificial intelligence and the Internet of Things, Contextual Awareness shall be the North Star which all these segments will look up to.

Sign up for our monthly mailing list