Quote:
Originally Posted by bob_ninja
Say I am interested in cyber fiction. I may hear of few big stars in the genre from media, I may stumble upon one of his books in Chapters store, etc. I only get exposure to big starts of the genre.
Say we had a number of forums here, one for each genre. Now I can simply go to this forums and examine the recommendations. You can even assign multiple tags, like dark cyber, or techno cyber, or whatever. Using indexing engine I could quickly search huge library of recommendations and fine tune for my taste or mood. Again, so long as it is an open forum not controlled by commercial interests then all authors have the same treatment; as opposed to what publishers decide to promote, or media decides to write about.
|
The problem I see is that sometimes it's hard to relate what you like to a label, or that you're interested in a specific subset of a labelled category. The examples above, for example, don't really man much to me--I know what cyberpunk is of course, and based on the names I suspect I'd dislike the "dark" and like the "techno" subgeneres. I'm sure there's dozens of categories that exist, and it's probably too much to expect a general reader to learn what all of these mean.
Probably there's a tutorial somewhere out there that defines each, and gives example authors and titles of each. But what if you have't read any of the books or authors given as examples? What if the authors cited as exemplars wrote in several subgenres, and the reader has only read the "wrong" books? A description of each classification will also be of most help if the reader understands their own personal preferences, something that may only come from reading a wide range of material, and having a few disappointments to define the limits.
So the question is, should readers need to educate themselves before deciding which group(s) to patronize? This could be a high bar to set for many readers, unless this tutorial is very carefully designed and executed. The problem is that reading dry descriptions simply does not work unless the reader has already has encountered material that can be related to the description. However including a "reading list" of plot outlines, short stories, and excerpts would turn this into more of a literature course, so is probably not the way to go either.
All this of course leaves aside the question of whether the book/author in question is even correctly catagorized. Certainly publishers tend to link books to whatever category sells best right now.
It is also probable that a reader's tastes are not limited to a single genre or subgenre. I suspect that a large part of what a reader likes could be determined by 20 or less rating scales (statistical dimensions) such as:
- how important is technology in the story or background?
- do you prefer characters or or noncharacter (plot, technology, cultural) elements to dominate he story?
- scales: personal or cosmic?
- involved, multilayer plot or action?
- hero or antihereo? Can you tolerate unhappy endings?
and so on. For these questions to have more than academic meanings, you'd have to relate them to one or more authors/stories that the reader has actually read. This could be a problem for a beginning reader.
Once ou had those dimensions for a book or author, you could almost certainly match readers with the material they'd best like. Each work would be rated by multiple readers on each dimensions as well as a few global ratings that even untrained reader could easily give (how "good" was it? did it flow well? would you read more like it?). These global scores, combined with the closeness of the match to your preferences, would give a personalized book score.
There are probably a lot of ways to determine the important dimensions, both to identify the questions that best measure them and to determine the scores for each work. It is very important to find the right set of measures or the whole thing is inefficient (less accurate, too many questions required to categorize, or too big a dtabase needed). There are statistical ways to compare how good a scoring system is but I'd suspect you might need tens of thousands (more likely hundreds of thousands) of whaterer data points (book reviews, user ratings, or even favorite book lists) to determine the best set. Maybe this has already been collected somewhere. I know Amazon has a huge database of what people buy, but whatever they're doing with it it isn't working too well, I find lass that 5% of their recommendation even worth clicking on the link, and some are simply ludicrous.
.