The document discusses content-based recommender systems, emphasizing collaborative filtering and user/item feature modeling. It covers the advantages and disadvantages of different recommendation approaches, including practical examples from LinkedIn and Twitter. The conclusion highlights the effectiveness of collaborative filtering while noting the importance of leveraging specific user and item features in industry applications.

1. Content-based
Recommender Systems
jake@ twitter.com @pbrane
User Interest Modeling, Twitter Inc.
Apache Mahout PMC
(previously: LinkedIn, a bunch of tiny startups)

2. Overview
• Collaborative Filtering == RecSys?
• User/Item content and/or metadata
• RecSys training w/ user/item features
• Advantages / Disadvantages
• Historical production examples from
Twitter and LinkedIn

3. Recommender System
“traditional” RecSys

4. Aside: “users” don’t
have to be users
• At LinkedIn, the Recommender Systems
team built a general-purpose entity-toentity RecSys: Product [user, item]
•
•
•
•
TalentMatch [job posting, user-profile]
GroupsYouMayLike [user, group]
{Jobs for your group} [group, job-posting]
AdsYouMayBeInterestedIn [user, ad]
Anmol Bhasin, Monica Rogati (now VP of Data at Jawbone), and
myself built...
So how do recommender systems work?
next page: “an artists depiction of collaborative filtering”

5. Collaborative Filtering

6. CF is Generic!
• users / items reduced to GUIDs, could be
anything
• large body of acad. work on techniques:
•
•
SVD, ALS + other matrix factorizations
stacked RBM, etc.
• General purpose OSS CF recommender:
• Apache Mahout (http://mahout.apache.org)

7. What about Domain
Specific Knowledge?
• Items are more than just GUIDs
• Users are more than just account names
• Perhaps, they are both much more:
• user profile text on Facebook / LinkedIn
• webdoc content + metadata
• movie genres, directors, description
• ad landing page content
could be derived data: at Twitter, every piece of text content
passing through the system gets classified into topical categories,
and users get classified according to their topical interests and
things they’re “known for”

8. User/Item Features
• each user has a feature-vector
• each item has a feature-vector
• dimensionalities may (will!) differ
• collectively, we thus have MOAR Matrices!
next page: more art!

9. Feature Matrices
we could decompose this resultant
user/item-feature matrix...
slight misrepresentation: user-features along rows of first matrix,
columns are user-ids
note: the “multiplication” here could be actual matrix mult, OR
maybe a more bayesian / statistical form: p(user|user-feature),
p(item|user), p(item-feature|item) -> p(positive engagement |
user-features, item-features). Full joint distribution ->HARD.
Naive Bayes? or...

10. Train a ranker/classifier
• take a column of user-feature matrix:
• row of item-feature matrix:
in
• embed
• train classifier to predict ratings given
go back and forth on this page to the previous one
next page is some notes about this

11. Training, cont.
• note: no need for any relationship between
features
• if you apply a discretization technique, don’t
even need to care about correlation
between +/- values and “goodness/badness”

12. Classifier/Ranker
RecSys HOWTO
• incoming preferences are triples of
(user-feature vector, item-feature vector,
preference-value)
• train classifier (online if desired!), and
• trained classifier spits out predicted rating
given user/item pairs
• note: may require some item preselection
predicted ratings may not be what you want, it may be a Learn To
Rank setup

13. Variations
• What if your features have some structure?

14. Structured data
• Item = { field1, field2, field3, ... }
• User = { fieldA, fieldB, ... }
• field1: tf-idf-weighted “position description”
• field2 : standardized categorical job title
• field3 : #years experience
• fieldA : tf-idf “job requirements”
• fieldB : #years of experience required
note: this is TalentMatch: here “items” are LI profiles, and “users”
are job postings

15. Pairwise-field similarity
• Some fields are naturally comparable to
others, can compute vector cosine, jaccard,
etc.
• Others have a business-specific similarity
f(#years experience - required experience)
• Each set of field pairs generates an
untrained weight

16. Train a low-dimensional
classifier/ranker
• take these O(|item_fields| x |user_fields|)
weights and feed into the training of a
ranker.
• given low number of features, very
interpretable
interpretation: p(user is good for job) = w_jobtitle+headline *
sim(jobtitle, headline) + w_jobdesc+headline * sim(jobdesc,
headline) + w_jobdesc+currentdesc * sim(jobdesc, currentdesc)
+ ...

17. Content-based RecSys:
Pros
• Fixes cold-start problem
• Scales fantastically
• Flexible: can Learn To Rank using LR, SVM,
GBDT, whatever
other content approach to cold-start: unsupervised similarity to
engaged-with items/users + CF
scales: use as much data as your classifier/ranker needs to
converge well. once trained, can often be a very low latency
method of generating item scores. Many classifiers are extremely
insensitive to #features input

18. Content-based RecSys:
Cons
• Not always very general (although: pairwise
crossed features are pretty general)
• Features may be too coarse
• Feature selection may be difficult
• Low-latency from large item sets is hard
• Underweights popularity, similarity to
known good items
for item selection: clustering can’t always work very well, if using
crossed features, but sometimes tricks like LSH can help

19. Hybrid Models
• Classifiers/Regression models yield scores
• can combine this score with preferences
from CF
• alternately, generate top-K items via CF,
rerank with your content-based ranker
(using CF rank as another feature)

20. Examples
• LinkedIn’s original (2010) generic entity-toentity RecSys was primarily content-based
• Twitter’s #discover product is a hybrid
recommender with content, social, and CF
features
Note: PYMK is not primarily content based
Also: personalized search is naturally a hybrid content-based
recommender

21. Conclusion
• Free paper title: “On the unreasonable
effectiveness of CF on the consumer web”
• But if you do know features about users/
items: learn to rank using them!
• This is more common than you might
think, in industry. But everyone’s got
different domain-specific features, so less
research about it
CF works absurdly well, given how little it knows about the items
it’s recommending
Riff on Hardy’s “On the unreasonable effectiveness of mathematics
in the physical sciences...”

22. Questions?
jake@twitter.com
@pbrane
LinkedIn/G+ : jakemannix