Is BookLamp really solving the right problem?

[Ricardo Cabral]

Do people really want to know which books were written the exact same way?

Really, what's the point of reading a book very similar (even in written style) to one you've already read?

Wouldn't they be interested on being recommended new books not exactly similar in style, length etc but with a high degree of correlation with what other similar users have been reading?

[Andy]

All this has been talked about, It will be possible to change what you are looking for and get something similar but not exactly the same.

[Aaron Stanton]

This has been discussed before, in depth, but I'll go ahead and address it again really quickly, because there's a fairly easy answer that's sort of one of those, "Oh, of course."  The simple answer is that people already seek out content that's similar to what they've read before in terms of writing style - that's why people have a tendency to read the entire back catalog of a single author.

I like Michael Crichton, and have read nearly all of his books.  These books all have matching style, but are different, and I read them because I could rely on the style of the next one to be a reasonable gateway to the interesting story.  Everyone has an author that they like more than others - most likely, the writing styles between that author's books probably remain fairly constant, with some level of variation, but we are still very loyal to individual authors.

Which is a fairly obvious answer to your question of whether or not people seek out similar styles.  They do, and it does work.  The problem comes when you exhaust the obvious - I've run out of Michael Crichton.  Now I can either look for a single author whose back catalog can be consumed, or - hopefully - I can turn to BookLamp to find a multitude of authors that have a similar style, but with probably more diverse storyline ideas, since they're from multiple people.

Seeking out similarities in writing style does not preclude the ability to find varying storylines and plots, or new ideas.  It just means that, when you seek out those new plots, you won't be impeded by the writing style.  Conceptually, it's important to realize that writing style can either be a positive or negative element of a book, and impacts enjoyment, but doesn't replace the importance of the other two elements, which are storyline and characters.  Ideally, writing style is invisible to the reading experience - but I've put down a lot of books without finishing because I hated the style, and because there were thousands of books out there with styles I enjoyed and with equally interesting storylines.

Also, you make an assumption that the ideal match is one that is exactly the same, which isn't the case.  Part of the self-learning aspect of the system is to determine not only what styles you prefer, but your preferred amount of variation from that baseline - for many people, the value isn't in having an exact match, but having a match that is similar enough not to put them off, but different enough to be new.

I hope that answers your question.  And Andy was also correct - the system also has the capability to reverse that approach - you can select a book and have it find you new books that are varying degrees of different from the source material, if that's your preference.

Aaron

[Stephen Rollins]

The simple answer is that people already seek out content that's similar to what they've read before in terms of writing style - that's why people have a tendency to read the entire back catalog of a single author.

I've always wondered, ever since I first heard of BookLamp...  What if you don't necessarily read the books for the writing style?  For example, I've read almost all of Anne McCaffrey's Pern series, but I've never take the time to read any of her others.  I like the writing style, perhaps, but what I'm really interested in is the story.  In that light BookLamp wouldn't necessarily serve the purpose it's meant to serve, because it'll refer me to books with similar writing styles while I'm really looking for books with similar stories or topics.

[Aaron Stanton]

I certainly agree - in the long run, the plots behind a book are more important to me than the writing style alone, but BookLamp's ultimate goal is to offer recommendations that match both storyline and writing style.  Keep in mind that many of the items your talking about looking for - similar stories and topic - can be easily searched via meta data.  You can and do search through genre, etc.  Very important stuff - our goal is to let you search within that same meta data for books that have all of the right elements.

In many ways it's a matter of numbers - if you think about the number of books that are out there, and the number of books you read in a year, it's almost certain that there are thousands of books out there you would probably consider good that you'll never get around to reading.  So while Book A might have a good storyline, good enough that you'll slug your way through to find out what happens even though the author is too wordy for no reason, it's likely that there are ten other books with equally good storylines, equally interesting characters, but with a writing style that would be invisible to you, instead of something that gets in the way.

So, you might enjoy Book A because of its storyline, in spite of the writing style, but you'll probably like Book B better, because of its storyline and writing style.  Our goal isn't just to build a recommendation system, but a more complete, well rounded recommendation system by tracking as many elements as possible.  Writing style is just one that we track in far more detail than other systems (that I know of).

Let me connect that to a personal habit of mine - maybe you'll connect with it, maybe you won't (I may be the only person that does this).  When I go to the library, I never leave with one book.  I almost always leave with 5 or 6 - each one carefully selected because the storyline described on the back cover sounds interesting.  Now, I know that I won't have time to read all six before I return them, but I get six because I assume that at least four of those won't hold my interest - I'll start them, only to find that the writing style just doesn't do it for me and I'll move on to the next.

Basically, my current system of finding a new book to read assumes that - even though the premise of the book is interesting - there's a good chance I will pass it up in favor of one of the other books that suit me better.  BookLamp ideally aims to improve the process some by trying to make sure that all the books in the stack will hold your interest, freeing you to worry exclusively about how interesting the plot is.

Hopefully, BookLamp frees you up to worry only about storyline, not about whether or not you're going to open the book and put it down after 30 pages because you just don't like the way it's written.

At least that's the idea.    As I said, it's possible I'm the only one that has this issue, but I don't think so.

Aaron

[Therin]

Phew!  It is so good to know I'm not the only one.   

[Stephen Rollins]

I certainly agree - in the long run, the plots behind a book are more important to me than the writing style alone, but BookLamp's ultimate goal is to offer recommendations that match both storyline and writing style.  Keep in mind that many of the items your talking about looking for - similar stories and topic - can be easily searched via meta data.  You can and do search through genre, etc.  Very important stuff - our goal is to let you search within that same meta data for books that have all of the right elements.

Oh, I didn't know that BookLamp was going to look at anything other than the writing style...  Mostly because I'm not sure how the system could possibly analyze the story of the books.  Would it search for keywords?  And what about genre, does that have to be inputted by the people at BookLamp, or is there some way the computer can figure it out, too?

[Aaron Stanton]

Yes, the ultimate embodiment of BookLamp is a merging of social input and objective analysis in such a way that BookLamp retains the objective nature of the objective analysis, but also gains the strengths of social recommendation systems.  Through scene-by-scene writing style analysis we can do things that other systems simply can't, but since both systems have different aspects that they do better than the other, the goal is to create a blended system.

Ultimately, it's easy for us to incorporate those elements on top of the objective Pandora-style analysis, and having the ability to use that data if we feel it improves the results is always a positive option.

We're also working on automated ways of tracking things like storyline and such, but much of the metadata, such as genre, is already available via the publisher or some similar system, so that's easily included.  We just don't allow you to filter through genre at the moment because of the size of the book database, and because most of the books are all the same genre.

Aaron