Where do we take it from here?

The Personal Digital Index will be extended in the following areas:

• Personalisation
• Categorisation
• Data Inputs

Personalisation

We intend to extend personalisation by taking the search terms and translating between semantically equivalent alternatives. When A says to B “That’s wicked…” he can mean it’s either a bad thing (as in nasty) or be using vernacular to indicate that it’s good. It depends not only on the speaker but also the context. We call this “semantic stemming”.

Importantly, we see the application of semantic stemming in the European language versions of the product. Many people have a basic reading ability in another language. Imagine if you were searching for the term in English and it also found documents in French and German that referred to the same term?

The Personal Digital Index provides a perfect base for this type of semantic processing with its frame of reference being the individual who owns the Personal Digital Index.

Categorisation

In order to speed up the search process, we will develop an automatic, intelligent categorisation engine that will further assist users in organising their information.

Data Inputs

We intend extending the data inputs for the Personal Digital Index as new technology comes on stream. In due course, we intend to make the following searchable:

Images

Imagine being able to make digital photographs part of your Personal Digital Index. You could then share these images with the people in the pictures automatically.

Our research has shown that commercially available facial image recognition algorithms are remarkably good as long as the set of people you’re trying to recognise is relatively small (< 250 people or so) and the individual in the picture is facing the camera!

Video

Theoretically the image recognition technology described above can be applied to a video stream (which is just rapid processing of a collection of stills). However we do not believe this technology will be generally available in the near future.

Sound

Although not high on our priority list as yet, if the voice to text recognition software gains a broader appeal then we can treat it as just another data source to the Personal Digital Index.

Interestingly, early Beta versions of Office 10 (Microsoft's next release) include some voice-to-text capability built-in.