How The System Does It

We say two Bible verses are linked when verse A and verse B are related to each other. Williams Bible Links™ collects Bible verse links cited in public domain works such as Bible commentaries and cross-reference collections.

Links are identified by seeing that two verses are part of the same topic, a group of verses associated together in a source. A footnoted concept in a verse, such as "prayer," and its cross-references are a topic. If a topic has, say, three verses -- the original verse and two cross-references -- then there are three links: A link between the original verse and the first cross-reference, a link between the original verse and the second cross-reference, and a link between the two cross-references themselves. The number of links goes up rapidly as topics contain more verses.

The system tallies how many citations each link has in our sources. This is a measure of how much consensus there is that the link is important. The more citations, the more authors thought the relationship between two verses to be interesting.

The abridged version displays the top-cited references for each verse.

The unabridged version divides the references for each verse into levels according to citation count, with level 1 containing the top-cited references, level 2 containing the next tier down, and so on. Sometimes a level is empty. That is normal, as level 1 appropriates references from lower-tier levels if necessary in order to make a full page.

The system only uses links "confirmed" with two or more citations. This is to minimize false positives -- verses that are not valid cross-references.

The system contains one million "confirmed" links (with two or more citations). Each link results in two cross-reference entries: Lookup verse A points to cross-reference B, and lookup verse B points to cross-reference A. So the unabridged version actually has two million cross-references. The abridged version has close to half a million.

The system attempts to group each lookup verse's references into topics, using network community detection techniques.

Rodger Williams


Public Domain