Sign in with Facebook
Recommendations And Discovering Good Reads
Many of you have given us feedback that you'd like to see book recommendations on Goodreads from a Netflix-style algorithm. We've always agreed, but good recommendation algorithms are really (really!) difficult to do right. We built Goodreads so that you could find new books based on what your friends are reading, and now we want to take the next step to make that process even more fruitful.
To tackle this highly complex challenge, Goodreads has acquired a company by the name of Discovereads.com. With their deep algorithmic book recommendation technology, we’re going to be able plumb our database of 100 million book ratings from 4.6 million users to find general patterns of the kinds of books people read and to generate high-quality personalized recommendations.
Book discoverability has become an increasingly difficult problem for readers. More than 300,000 books will be published this year, and the rise of ebooks and self-published books has made the market even larger. It’s a really, really, hard problem to get right, and we we’re excited to have the Discovereads team onboard, because along with being great people, they have a background in machine learning algorithms.
Inspired by the Netflix prize (and with some direct help from a member of second place team), Discovereads’ Brian Percival, left his PhD at Stanford University in Electrical Engineering (in neural circuitry software modeling) to pursue this project. Together with CEO Kyusik Chung, the company has taken three years to refine the technology.
Discovereads combines multiple machine learning algorithms and graph analysis techniques to analyze book ratings and spot trends. Its technology will also improve our unique targeted advertising program. Currently, authors and publishers can purchase ads for their books and target the ads to members who have read similar genres. Hopefully, the new technology will make it more likely that when you see a sponsored book, it will be even more likely it’s something that you’re interested in!
We expect to have Discovereads fully integrated within the next few months. And as always, when it goes live, we will welcome your feedback!
To tackle this highly complex challenge, Goodreads has acquired a company by the name of Discovereads.com. With their deep algorithmic book recommendation technology, we’re going to be able plumb our database of 100 million book ratings from 4.6 million users to find general patterns of the kinds of books people read and to generate high-quality personalized recommendations.
Book discoverability has become an increasingly difficult problem for readers. More than 300,000 books will be published this year, and the rise of ebooks and self-published books has made the market even larger. It’s a really, really, hard problem to get right, and we we’re excited to have the Discovereads team onboard, because along with being great people, they have a background in machine learning algorithms.
Inspired by the Netflix prize (and with some direct help from a member of second place team), Discovereads’ Brian Percival, left his PhD at Stanford University in Electrical Engineering (in neural circuitry software modeling) to pursue this project. Together with CEO Kyusik Chung, the company has taken three years to refine the technology.
Discovereads combines multiple machine learning algorithms and graph analysis techniques to analyze book ratings and spot trends. Its technology will also improve our unique targeted advertising program. Currently, authors and publishers can purchase ads for their books and target the ads to members who have read similar genres. Hopefully, the new technology will make it more likely that when you see a sponsored book, it will be even more likely it’s something that you’re interested in!
We expect to have Discovereads fully integrated within the next few months. And as always, when it goes live, we will welcome your feedback!
Comments Showing 51-84 of 84 (84 new)
date
newest »
newest »
message 51:
by
Lyn (Readinghearts)
(new)
Apr 08, 2011 02:50PM
This looks both good, and deadly to my TBR list!
flag
I am so excited about this. I can't wait to see it up and running. Not that I don't already have like 400 books on my to read list...(:
I was just thinking that something like this was needed. I find myself added "to-be-read" more and more, but I really have no idea if I'll like the book or not. And sometimes I'll discover a really really great book through one of the site members, but then don't know the first place to look for more like it. Word of mouth Recommends only work if other people know the books exist. And I don't have enough real friends who I'm tracking on here to know when I've been exposed to a great read.
Excellent! Im very excited about this! I had just emailed you about being able to do this and Im happy to see that you'll be doing it!!
Yay! I also have wished that Goodreads had recommendations at the level of Discoverreads.
I have been waiting for something like this! Amazon recommendations are laughable and even those from library sites are pretty predictable. The question is always why we liked a book, which isn't obvious even to us. Was it the setting (I'll read anything that takes place in Paris), the time period, the characters, the tone, the plot, the writing itself? Along with Netflix, I'm thinking of Pandora, which breaks down a song into elements of rhythm, orchestration, key, etc. But it's so much more complex to do that with a book of 300 pages than with a 3 minute song or even a 2 hour movie.
What if we have to start paying up for the service? :O
You guys rock. I was just saying to myself the other day that this was the only thing the site lacks.
That's great! I love book recommendations Netflix-style, can wait to see.
As a new user I was recently wondering if that would ever happen - glad to hear it!
Sounds pretty awesome. I'll have to load in the rest of the hundreds of books in my library so I can have more accurate recommendations.
Danielle wrote: "I think that this is wonderful for quickly seeing what is available that we may not have heard about... there are a lot of books out there that deserve attention but haven't got the buck for the pr..."I would love it if it showed readers with similar interests as me as well! Great idea!
There are a number of features that predictive analytics can include that I'd love to see.If you want to see any of these features, or suggest others, comment on this comment indicating the feature number you also want, or with other related features I haven't mentioned.
1. OKCupid's system defines several personality characteristic pairs, displaying them as two-dimensional spectrums graphed as normal distributions; the more information a user gives the system, the more pronounced the peak and certain the description, according to OKC. I don't know if they still do this, but in the past I could look at the characteristics that OKC's system had assigned to me, and see those graphs. It was really interesting and made the geek in me super-happy. It also helped me understand the matches the system would give me.
2. A few people have mentioned the ability to browse other users based on the similarity of their preferences. That would be great.
3. If #2 could be refined further to allow us to say, "I want to find books/users that match my preferences not including A," or "including only the following aspects of my preferences."
4. Pandora will tell you why it played something on a given station. Likewise being able to see why your system recommended a book, or thought a user had similar preferences, would be valuable.
Please tell me this thing can be opted out of...Sure, it's good to learn of new books- when you're looking- but I thought that was why we're here. At least, that's why I was here.
If i'm going to have the computer tell me what I like without being able to shut it off, then i'm outta here.
Besides, for all assurances to the contrary, maybe it's just another book of the month club.
Pope wrote: "Please tell me this thing can be opted out of...Sure, it's good to learn of new books- when you're looking- but I thought that was why we're here. At least, that's why I was here.
If i'm going ..."
Don't we already have sponsored books? One would think you would have a bigger problem with that....
Danielle wrote: "I am curious however if it would be possible to add in a feature to the compare bookshelf feature between users to show those users that have the highest rate of compatibility to you instead of having to wade through random users?"I also want this feature.
This sounds really interesting.However, will it have book recommendations flooding the screen each time you log on? or will it be quite unobtrusive in that as a user you have to click on a Goodreads tab to find the recommendations?
We'll find out soon enough...
Very clever, most of the time I find it very hard to get a good book with the genres I like. Hopefully this will help! I really can't wait to try it out!
Hip, hip, hooray!!! Thank you, thank you, thank you!When I first started using Goodreads, I looked for a magical tab for recommendations--geez goodreads had everything else. For awhile I thought I missed it.
Unfortunately now, I'll just need to buy more books...... Launch date?
very glad to hear of this upcoming feature. And yes, I vote for something that provides good book recommendations and can identify highly-compatible book shelves and other readers of like interests.
I am very new to Goodreads and did not know about the recommendation feature....how does one sign up to to have access to these recommendations and reviews?thank you for great information
Frieda wrote: "I am very new to Goodreads and did not know about the recommendation feature....how does one sign up to to have access to these recommendations and reviews?thank you for great information"
This is a feature that is going to be new and isn't available yet. Hopefully it won't be very much longer as we have all been waiting with much excitement!!!!
This may add something. Matching overlaps within clusters of books read by individuals and then recommending those books outside the overlaps plus providing a look inside capability would be basic. it's like visiting a book store. But beyond that?? How do we discover new authors and new subjects of interest? Work is required. You have to look inside books with your own two eyes...and let your brain exercise. We shouldn't get too lazy.
So... almost two months since this announcement and no new information. :( Anyone know how this is moving forwards? (I have family who really badly want to try it out, and it's the only way I'll ever get them on GR! xD)
Karen wrote: "Hope this helps me. I like to read novels set in another country where I can both care about character and learn something new about culture or geography. Have been going to groups that have coun..."Please consider joining "Fans of Maps" - we're trying to read around the world learning about the culture and geography of various countries and regions this season.
(I'd have sent you a Private Message to answer your query, but you're not accepting PMs.)
Kinga wrote: "So.. can we get any update on this????"See the thread posted in the Feedback group yesterday to help test some of the preliminary results.
Rivka, I can't locate that thread, could please post/send me a link?
You have no recommendations based on your bookshelves yet. <---- i have been stuck with this ever since i messed around with my shelves, any suggestions ?
How come I can't simply get recommendations on books a bunch of my friends have read and liked?
