Union of Intelligible Associations
Union of Intelligible Associations

Projects Overview (Explanations)
Global Strategies Project (Explanations)

Method: Computer-assisted generation of strategies

Global Strategies Project


This Encyclopedia is based on the assumption that there is an intimate relationship between its different sections and with the international organization profiles in the companion Yearbook of International Organizations. For this reason, ways were sought to use experimental computer programmes to scan such texts "intelligently" and to generate potential strategy names from existing formulations. These potential strategies could then be reviewed for inclusion and/or subject to manual editorial refinement. There were two principal initiatives of this kind.

1. From human values and problems

As described in the Human Values section of the Encyclopedia (Section VZ, Volume 2), there is a clear link between what are there identified as "destructive values" and well-formed names of "world problems". The problematic nature of a problem only becomes evident when its name includes an appropriate negative value. Thus for the World Problems section (Section P, Volume 1), "youth" or "peace" are not problems as such. Whereas "alienated youth" or "unjust peace" may be so considered. This explicit link between negative values and problem names enabled specific links to be established between the negative (destructive) values enumerated in the Encyclopedia (Section VD) and the corresponding problems (Section P), as can be seen from the cross-referencing in the entries in Section VD (Volume 2).

It is apparent that many strategies (or action proposals, solutions, etc) are formulated as direct responses to perceived problems. Based on this observation, the problem names clustered under each negative value in Section VD were presented for editorial inspection as candidates for conversion into strategy names. For example, the problem "Complicated spelling" was converted into the potential strategy title "Simplifying spelling".

The problem "Substandard housing and accommodation" produced the strategy "Raising standard of housing and accommodation"; it also generated a tentative strategy "Specifying basic housing standards", which was later subtitled "Developing housing codes". Similarly, problems concerned with substandard and defective consumer products implied strategies concerned with "Improving product quality control" and "Strengthening consumer rights".

The negative notion in a problem name might be retained to produce a "negative" strategy. For example the problem "Self-deception" was converted into the strategy "Deceiving", subtitled "Using deceit". Alternatively the gerund at the beginning of the strategy title was created directly from the "negative" value; thus the problem "Criminalization of prostitution" became the strategy "Criminalizing prostitution".

Existing strategies relevant to each problem cluster were presented within the same framework during this process, in order to facilitate editorial judgements. This approach made it possible to establish specific links between problem entries and candidate strategy names. Thus the problem "Criminalization of prostitution", referred to above, might be also cross-referenced to strategies previously derived from other sources, such as: "Decriminalizing prostitution", "Legalizing activities", "Correcting criminal stereotypes", "Studying criminal subculture" and "Harmonizing criminal law".

Specially coded, such tentative strategies were then fed into the strategy database where they could be confronted with duplicate or similarly named strategies. This then permitted editorial merging of entries, preserving the links to specific problems.

Approximately 15,000 potential strategy names were originally formulated in this way. Editorial reduction and compression has to this point reduced the number by around 5,000. Most of the remaining entries are insufficiently worked (lacking text and/or sufficient additional cross-references) and are currently ascribed to the SG, SJ or SK sections (not published in this volume except by cross-reference in one of the published strategy entries).

2. From international organizations

It can be assumed that the declared aims and activities of international organizations would state or imply the strategies or actions undertaken by the organization. Indeed, these may even be implied by the title of the organization. The profiles of international organizations in the Yearbook of International Organizations are therefore potentially a rich source of information on strategies.

The logistical challenge is how to extract such potential links between strategies and the organizations concerned with their use from some 30,000 organization profiles in the database without inspecting each profile individually. The approach taken was to design a computer programme to scan the title, aims and activities of the organizations for phrases implying a strategy. The resulting 161,267 candidate names were then individually scanned, reducing them to 37,826. This excluded many improbable strategy names. The remainder were then written into a file of candidate strategy names, linked in each case to the source organization.

The candidate names were then subjected to several processes. In a first pass, names matching any in the strategies database were selected out. These numbered 4,633 for which the organization cross-references were transferred into the corresponding entry in the strategies database. A further 13,000 new strategy names were then added to the database with their organization links; after combining duplicates this effectively made some 7,000 new strategy entries.

Subsequent editorial work has discovered that a portion of these potential new strategies in fact duplicate existing strategies with alternative wording. In such cases the duplicates have been combined, thus enriching the alternative forms of the strategy titles and preserving the cross-reference to the source organization. Prior to closing for press, all strategies derived from organization entries which had not been linked to other strategy entries were transferred into Section SK.

Despite the assistance of the computer in generating the names and relationships (an enormously time-consuming task if done manually), individualized attention must still be given to verifying that a strategy and organization are indeed matching. For example, the computer may have generated a credible strategy from an organization's text, but which in fact is not the concern of that organization.

For example, the strategy "Researching work" could be undertaken by trade union or employer organization but would be regarded as a suspect cross reference to a scientific research organization undertaking research work. Plausible matches are the most tricky to detect as errors and require reference to the original organization text.

Also requiring further work are the situations in which the text of one organization generated a relevant strategy, but the text of another organization employing the same strategy did not because the descriptive text did not allow the strategy to be isolated. The mix of organizations associated with a given strategy, therefore, provides useful examples, but may in many instances require further attention. Strategies with which an organization is associated may not be adequately identified in a general profile of that organization as it appears in the Yearbook of International Organizations, or in its own documents for that matter.

There remain, therefore, rich opportunities for work with the thousands of candidate strategy names derived from organizational material and as yet not used.

From Encyclopedia of World Problems and Human Potential