CORPUS THOMISTICUM
Enrique Alarcón & Jaime García Hoz
Chartae synopticae operum S. Thomae de Aquino
automato IBM DB2 Intelligent Miner statistice exaratae
Pompaelone ad Universitatis Studiorum Navarrensis aedes A.D. MMXIII

We present here 79 synoptic maps of writings by Thomas Aquinas, and 42 of other related works.

These diagrams are the result of a statistical analysis produced using DB2 Intelligent Miner, donated by IBM through the IBM Academic Initiative.

We have applied a statistical procedure called Affinity analysis: mathematical algorithms applied to large databases to discover and to evaluate which elements are associated, and their mutual relations. These rules of association can be represented graphically, so that the colors and the size of the connecting arrows express the various statistical valuations in an intuitive manner.

Affinity analysis is applied here to frequent and meaningful words (lemmata) in the Index Thomisticus database that co-occur in sentences of single works. For indeed, all texts are formed of grammatical sentences that connect words; and when, in a work, two terms co-occur frequently in the same sentences, it is probable that they express a reiterated, and therefore important issue. As a result, the rules of association between the terms of a text give an idea of its frequently-occurring contents; and the corresponding graphic representation shows, as in a synoptic map, the main topics of the text analyzed.

We are here moving in the realm of probability and, therefore, it is important to evaluate the results in the same light. A priori, the longer a work is, the lower the probability that its words will co-occur in the same sentences by mere chance. As a result, and from this perspective, the method employed here has a higher validity the longer each text is. For this reason, the length of each analyzed work, measured by its total number of words, is given underneath each synoptic map.

The frequency with which each of the associated terms occurs is also relevant. In fact, a high frequency probably indicates a greater relative importance of the topic at hand. In these synoptic maps, the most frequent terms appear in orange, infrequent terms in blue, and intermediate cases in white.

Another significant statistical value is the frequency with which, in each text, the same two words appear in a single sentence. Technically, this is measured via a proportion called support: the number of sentences that show this co-occurrence, divided by the total number of sentences in the work. In the graphics that we present here, the support is represented by the color of the arrows: orange when greater, blue if lesser and white for intermediate cases.

Finally, the reliability of each association between terms is measured by the confidence value: it is the proportion of occurrences of a given term where it occurs with another term it is associated with. Confidence is represented by the width of the arrows: the wider the arrow, the greater the confidence that the association rule has.