"The Prolog program was up and running before she went on leave"

Fungi are not just a load of old mushrooms and toadstools. If they were then perhaps identification would be easier. And for people who have a need to identify species accurately, it seems that LPA Prolog has been able to help make that task much less expensive and time consuming.

Ident is an application developed for the Forestry Commission aimed at cutting the cost of fungi identification. Cost is a problem because it is often necessary to grow the fungi in question for some days to find out what they are. Historically this task was done by the member of staff most familiar with the text key system in use at that time. Then, as now, fungi must first be grown in a culture to be able to analyse characteristics such as growth rate, and the type of development the specimen follows.

Tony Ludlow of the Forestry Commission, said: "A very common problem in identifying species, is that you know a subset of the information that would lead you to the answer. If you use a standard key in a text book, it begins with question one, if you can't answer question one you can't go on." This means lost time, and resources tied up cultivating the various specimens.

"In the studies that we have to do, it may take six weeks for some of the information to become available," said Ludlow. "If we can say we know that it's this species, or genus, after two days, then we don't have the expense of keeping it growing."

But the crunch came when the person responsible for identification had to go on maternity leave. As Ludlow recalls, once the need had been identified, making it happen was relatively straightforward. "The Prolog program was up and running before she went on leave."

Ident had originally started life as a classic production rule based system.The prototype was developed for the pathology branch at the Alice Holt Research Station to help classify the many samples of fungi it receives each year. Although the prototype proved very useful it was also rather slow. Ludlow's solution was to re-write the program so that fungi and their attributes were represented as sets.

In this way all the available information can be used as attributes that the sample does or does not possess. This makes it very quick to reduce the possible set to which the species belongs. Using LPA Prolog this was a relatively easy task that was accomplished in a matter of weeks, rather than months. Had the system been engineered in a lower level language such as C, Ludlow believes it would have been much more difficult.

See Next Application Brief