Reading Bob’s post and seeing the picture of Eli holding 37 volumes of the Church Fathers’ writings brings back memories.
I remember when we did the ECF project. That was Eli’s baby, and what an incredible job he did in pullling that thing together. I can recall helping his team out by proofing through Greek in footnotes to make sure it was actually correct. I think I may have even keyed some of the footnotes.
But what I remember most is the topical index.
Huh? You didn’t know that the 37-volume Logos Bible Software edition of the Early Church Fathers has a topic index? Well, it does. And that particular topic index doesn’t exist anywhere else.
This article looks back on how all that happened. Ahhhhhh … nostalgia!
It must’ve been 1996, but it sure doesn’t feel like it was that long ago. Eli and I were just figuring out what it meant to make electronic books, and were “kicking the tires” both with LLS resources and with our still-developing programming skills.
Bob, as is typical, challenged us. I began to think about how this big set of books we were working on could be still more useful as an electronic edition. It all coalesced one night. I wrote a fairly simple script that evaluated the text of the ECF volumes and determined topics for each article.
Pretty cool, I thought. But not cool enough.
Bob challenged us again. He pointed me to some books that discussed automatic topic generation, and used words like “document vectors”, “inverse document frequency” and all sorts of other weird stuff I didn’t understand. And math. My eyes glazed over. But I went with it, and made the topic generation better based on what I’d read, thanks to Bob’s encouragement … and patience.
Then Eli added his magic. He located a copy of a thesaurus from 1913 (or so), right about the time that the Schaff edition of ECF was published. He thought it would be cool to use that to generate even more entries based on synonyms.
We added the synonym stuff, culled through generated topics, developed stop lists … and ended up with a fairly decent topic index of the writings of the church fathers.
And it’s been in our edition of the Early Church Fathers ever since.
We’ve used the same code (or similar code) in other projects. Sometimes it is appropriate, sometimes it isn’t. If a book has a subject index already, we implement those entries as topics in the electronic resource. If it doesn’t have a subject index, and it’s the sort of book (or set of books) that might benefit from a cursory generated index, then we consider adding it.
This particular approach doesn’t always find the best or most appropriate topics, but if the content analyzed is suitable, the generated index is helpful. And it helps tie the resource in to other automated reports that rely on topic indexes.
Nine years ago, I’d never have anticipated that work would still be useful and helpful today. But that was just the start. We’ve learned a lot in the past nine years, but we also know that we have a whole lot more to learn. But what we do learn — we’ll try to apply it in ways that are beneficial to the study of God’s Word.
Just like that index to ECF we made nine years ago.