Now that we have covered basic techniques for recommender systems, choose one commercial recommender and describe how you think it works (content-based, collaborative filtering, etc). Does the technique deliver a good experience or are the recommendations off-target?
You may also choose one of the three non-personalized recommenders (below) we went over in class and describe the technique and which of the three you prefer to use.
1. Metacritic: How We Create the Metascore Magic
2. Rotten Tomatoes: About Rotten Tomatoes
3. IMDB: FAQ for IMDb Ratings
I rely heavily on Rotten Tomatos for movie/show consensus. While I am a registered user on Rotten Tomatos, the website has never “recommended” anything to me. I believe that Rotten Tomatos doesn’t have a recommender system. Rather it looks to be a review aggrigator which collects explicit reviews for critics and average users, categorizes and aggregates them. Movies and shows themselves have various features such as genre, actors, directors, awards, etc. With explicit review aggregation and movie/show categorization, Rotten Tomatos is able to slice/dice content in various features. Generally for me, Rotten Tomatos has worked well as a reliable gauge on a particular piece of content
Read the article below and consider how to handle attacks on recommender systems. Can you think of a similar example where a collective effort to alter the workings of content recommendations have been successful? How would you design a system to prevent this kind of abuse?
Travis M. Andrews, The Washington Post (2017): Wisdom of the crowd? IMDb users gang up on Christian Bale’s new movie before it even opens.
This kind of “hi-jacking” of the reviews to either artificially float or sink reviews is becoming more common. This not just realted to movies, this can apply to just about any context where a user opinion can be input. Product, and business reviews are another example where such “disinformation” can take place. Though there are some measures that systems can take to help alleviate this:
Analysis of reviews left: Usually fake reviews have a pattern and real reviews have a pattern. These patterns can be analyzed to determine if a review is authentic or a fake. For example, a fake review will have negative sentiment but may never mention anything material about the product that results in a negative sentiment.
User profiles: User profiles can be analyzed to see location and point of registration, in-conjunction of previous reviews (if any) and current review.
Rate of ratings/reviews: If an item is being flooded by reviews in a short period of time and the reviews are skewed in a direction then chances are there is something dubious going on
Timing of reviews: The product release date and review/rating time stamp can also be analyzed to determine authenticity
To spot unauthentic reviews all the above could be done in combination. Commerce platforms that host these products/items are also serious about fake reviews. This is because it can have a negative business impact. For example, if all reviews on Amazon cannot be relied on, chances are a user will make a purchase somewhere else or not purchase the item at all. Most customers today rely heavily on reviews as a facilitator for the user to complete the transaction on the said platform.