Peter Brown
Department of Computer Science, University of Exeter, Exeter EX4 4QF, UK
P.J.Brown@ex.ac.uk
We have a research programme that is concerned with retrieving information for users of cellphones and other mobile devices. We believe that one of its most promising uses is to help provide publicity for rural enterprises aimed at visitors.
People who run rural enterprises that are oriented at visitors tell us that publicising their existence is one of their biggest problems, particularly if they are just starting out. Leaflets may help, but are most appropriate for larger enterprises. A web presence helps, especially with accommodation and other items that people plan in advance; however, with millions of web pages in existence, awareness of any one page is a problem. Signs beside roads may also help, but are not always applicable. Above all there is a need for focus in publicity: reaching the people who are most likely to come. In addition there is a need for publicity to be flexible and easily updated. For example, though a big enterprise may have fixed opening hours throughout the year, the opening hours of a small enterprise may vary from week to week, depending on other circumstances. Moreover the nature of what is available may also vary from week to week. In addition the system should be usable for single community events that want to attract visitors too, e.g. the village Flower Show or church services; such events might be posted a few week's before their occurrence, and deleted afterwards.
We believe that there will soon be a facility that meets this need. This is the cellphone. Future cellphones will:
Indeed current cellphones already have the above facilities to a degree and there are already services such as `Tell me about hotels near by' or `Tell me about current road conditions'. The service we have in mind is `Tell me about the attractions nearby' and we are interested in a service catering for lots of small players, rather than, say, large hotel chains.
To use such services the user dials a certain number and the information they need is delivered, either in audio form or on the screen of their phone. The same services can potentially be used from computers rather than phones (in this case the location is not normally detected automatically, but the user specifies the location they are interested in, e.g. by pointing at a map). Here the larger screen allows delivery of more information.
Experience shows that, to be successful, such services must deliver information that is really relevant to the user. For example if the system delivers six pieces of information (`Here are the six that best meet your needs') and, to the user, only one of them is actually of interest, then the user will soon abandon the service as tedious and a waste of time.
Our research has been geared to solving this problem, i.e. in finding information that really is relevant to the user. We have produced a prototype implementation of a `context-aware' retrieval system, and this has proved successful. It would, however, need outside partners to turn it into a product.
Rural areas are a favourable application area, because there is less of a needle-in-a-haystack problem, e.g. there may only be four relevant locations within ten miles of the user whereas in the middle of a city there might be forty.
Assume the service was aimed at tourist attractions. A tourist wants to know about nearby attractions, They dial a certain fixed number on their cellphone, and get back the attractions that are most relevant (e.g. nearby, open, and suitable for the current conditions). The user does not need to enter their location and other aspects of their current context: the system knows this automatically; thus calls should be fast and convenient. The service should provide facilities for users to set their preferences in advance (e.g. gardens and plants).
When information is delivered, our software delivers a ranked list, with the most relevant, e.g. the closest, first.
Each tourist attraction supplies an item for the database used by the service. It is important that each item be readily changeable by the provider (e.g. in the coming week different opening hours, special display of fruit varieties). There is no reasonable why the system should be expensive to the information provider, though there is always the problem of the initial threshold: if the system had only a small number or users and providers it would be expensive to run, but if the system took off economies of scale would make it cheap for each user/provider.
Other services might be aimed at shops, such as farm shops, footpaths, Bed and Breakfast, etc.
Current disadvantages are:
We do not believe any of these disadvantages is a killer.
The huge advantage of this system is focus: information is only delivered to people who are nearby and have the right context. No other means of publicity comes close to reaching the right people so cheaply.