00:00So this ability to search on the concept rather than match keyword to keyword,
00:09does that contain some elements of artificial intelligence in it or is
00:13this something different? Yeah it's the same but only in the most naive sense.
00:18People who are specialists in artificial intelligence and machine learning will
00:22probably roll their eyes about this but what I'm doing is a very it's a type of
00:27what's called supervised machine learning. What is that? It's very simple actually.
00:31All that it means is that at every point we're presuming that researchers, humanists,
00:35want to be able to sort of go in and manipulate the results.
00:39So we're constantly tinkering with and playing with the results that we get in order to find things that are more useful.
00:45For instance using the search engine that I have you do a search and you'll get a list of not the 207,000 volumes
00:51but you'll get a list of the 13 million passages ranked in order of the likeliness of their usefulness to your project.
00:57Then you can go through and as you're reading if you find things that are particularly salient
01:01that you hadn't included in your loading set that you wanted to include in a subsequent search in order to modify that search,
01:07you can do that. You can go in and click on a few more passages and then select those terms that you want
01:13and at every point then you're supervising the learning of the search engine in order to arrive at a better result for you the researcher.
01:20When you were in this phase of thinking about these passages and how to analyze them
01:28and you came up with the idea of a different kind of search engine,
01:34I'd like to think that many scholars in your shoes would have said,
01:39oh I wish someone would develop this. So what do you think it takes to or what did it take you to actually think I need to develop this?
01:49Did you have a background that helped you get a part of the way? I mean I know you said you asked for help but there's a starting point there and I'm curious about yours.
01:59Yeah, so there's a couple things to say to that. I think there wasn't any single one starting point but there were a couple things.
02:06First is that I actually do have a bit of a background in the sciences. Even though I'm an English professor,
02:12I did most of my undergraduate training in biochemistry. I was not sure if I wanted to be a doctor, I was not terribly good at it.
02:19But it had me sort of thinking about quantitative methods and so that was kind of natural for me, that type of thinking I suppose.
02:25The other thing is that while I didn't work with them directly, while I was doing my PhD I was at Stanford and the literary lab at Stanford is one of the more important centers for this research around the world.
02:36So even though I wasn't directly involved in that myself, the ethos of it sort of like bled into my thinking about how one could use digital tools.
02:44So not only at the institutional level but also at the individual level. I had a couple of really good conversations.
02:49I wasn't exactly sure how to achieve this and I had a conversation with a colleague of mine named Mark LG Hewitt who's an assistant prof now at Stanford.
02:57And he had actually written his dissertation about the very notion of concept search.
03:02And he was using a slightly smaller corpus of novels that he'd acquired in order to look at the changing sense of the sublime and places where it might crop up in the 18th and 19th century that you wouldn't necessarily expect.
03:16But essentially that process that he was using seemed directly relevant to my own research and the fact that he'd done it made it seem all the more possible for me to go out and give it a shot.
03:27So you actually got to see other contexts of people trying this.
03:31Yeah.
03:32To give you some confidence.
03:34I think that's really helpful, you know.
03:38I'm not saying that it was like smooth sailing by any means whatsoever.
03:42Throughout this entire project, you know, sometimes I've been working on this individually.
03:48That's been really frustrating and hard.
03:49And sometimes I've been working on it with lots of people, but, you know, people that don't necessarily see this as their primary research project.
03:56And so, you know, there's there's lots of moments where things sort of gallop forward in a couple of weeks and then come to a standstill for another period of time and then gallop forward again.
04:07It's hard to anticipate.
04:08And I think especially when you're sort of trying to negotiate a new project or a new way of thinking about your research in this way, I think it's good to be fairly patient with yourself.
