1   A Computer Knowledge Base

The knowledge base 'Atwima' was constructed to address constraints brought about by declining soil fertility because of shortening bush fallows. It represents local knowledge. Thereby, it may enable identification of requirements for agroforestry interventions in participatory technology development. Soil fertility management, soil classification, weed management and cocoa shade trees are the particular concerns identified by its authors.

The knowledge base comes from the people of Kyereyase and Gogoikrom, of the Atwima district of Ghana. The computer programme named KT5 was used in a compilation of unitary statements to expressing single assertions in a 'definite clause grammar'.

There are seven topic hierarchies in the knowledge base 'Atwima'. Sub-topics may contain numerous statements, as for example, 201 statements are given for the topic 'weeds'.

2   Boolean Searches

From a Boolean search of the knowledge base, 'Panicum maximum and Chromolaena odorata' (local names esra and acheampong, respectively), four knowledge statements by the farmers are found. The four statements are:

• An increase in density of acheampong causes an increase in shade intensity of acheampong.
• An increase in the shade intensity of acheampong causes increase in intensity of acheampong shading esra.
• An increase in the shade intensity of acheampong causes a decrease in growth rate of esra.
• A decrease in growth rate of esra causes a decrease in density of esra.

The farmers' knowledge about ecological interactions of the two plant species may be inferred from these statements. Thus, it may be concluded that Chromolaena odorata' controls the growth and spread of Panicum maximum through shading, where the two occur together.

Further computer searches of causal relationships concerning these four statements (shaded boxes in the diagram) and all other directly associated statements give the computer generated diagram in Appendix 1: Atima knowledge statements, 'Esra and Acheampong' - causal relationship.

• The boxes, or nodes, may represent objects, processes, attributes, values or actions of unitary statements.
• Each node may contain one or more knowledge statements.

The relevant knowledge statements associated with nodes in the diagram are given in Appendix 2: Knowledge statements concerning Esra and Acheampong.

From the diagram it can be seen that tree density and shade, soil type and soil fertility are implicated in the two plants' interactions. From the associated knowledge statements, it is known locally that both Panicum maximum and Chromolaena odorata colonise soils of reduced fertility. However, Chromolaena odorata' is also known to cause an increase in fertility of the soil after three years fallow, unlike esra which causes a decrease in soil fertility. Thus where Chromolaena odorata' development in fallow is strong it may help to control colonisation by Panicum maximum through soil fertility changes.

Both plants are known locally to have a low tolerance of tree shade. However, as it is known locally that the density of Panicum maximum increases where tree growth rate is slow, I assume that the shade condition favours Chromlaena odorata growth over the growth of Panicum maximum. This may include conditions where the two grow together in low fertility soils.

Afonywie (sandy loam) is known locally to favour the growth of Panicum maximum. I deduce the possibility that this is so because afonywie causes the density of trees to be low (and I infer thereupon, the growth rate of trees is low with afonywie).

3   Further Querying the Data

No statements are found under the search string 'esra and acheampong and weed crop interactions'. It appears that the interviewees were not explicitly asked about such knowledge. However, statements were found for the Boolean search strings 'weed management and esra', 'weed management and acheampong', 'weed control and esra', 'weed control and acheampong' (See Appendix 3: Knowledge Statements concerning Esra and Acheampong, and Weed Management). In addition, some knowledge statements found under Boolean search string 'esra and acheampong ' are useful in answering the question concerning which weed is more of a problem to the farmers.

From these knowledge statements, I see Panicum maximum decreases soil fertility, unlike Chromolaena odorata which enriches the soil. Thus, the presence of Panicum maximum defeats the purpose of fallow. It has no locally reported benefits. Furthermore, its presence is known to decrease rice yields. When found amongst crops, I assume Chromolaena odorata is unwelcome by farmers because its stump sprouts and seedlings are reported to cause unwanted shading to crops.

I note that Chromolaena odorata is thought locally to be harder to clear from fallow than esre. If the task of clearing it is given to one group, then it I assume it would be more of a problem to that group only. As noted previously, Chromolaena odorata appears to control the spread of Panicum maximum. Therefore from a purposeful point of view, I believe Panicum maximum is less welcome to farmers than Chromolaena odorata in fallow land.

4   Overview of Interviewees and Knowledge

It is possible to gain information regarding the interviewees according to demographic variables, by querying the database. For example, in the present database, Six Ashanti men over 35 years and seven Ashanti women over 35 years were interviewed for the knowledge base. Older Ashanti women know more about weeds than trees, it appears from the statements in the knowledge base. They offered 11 statements about trees and 56 statements about weeds. The assumption here is that they were equally able to make explicit knowledge of trees and weeds.

Note: Definite and indefinite articles have been added here for ease of reading. These articles were absent in the computer versions.

5   Appendices