Augmenting personalized recommendations with explanations is believed to improve users' trust, loyalty, satisfaction, and recommender's persuasiveness. We present a exible explanations framework for collaborative ltering recommender systems. Our algorithms utilizes item tags to automatically generate personalized explanations in a natural language format. Given a speci c user and a recommended item, the algorithm utilizes the user's personal information as well as global information (e.g., item similarities, metadata) in order to rank item tags based on their \explanatory power". The top tags are chosen to construct a personalized explanation sentence which helps shed light on the underlying recommender. Our system has been well received by both focus groups as well as in expert evaluations and is scheduled to be evaluated in an online experiment.
BACKGROUND AND INTRODUCTION
In the context of recommender systems, an explanation
is a short textual description accompanying each
recommended item to explain the recommender's choice for the
user. The importance of such explanations for real world
applications is increasingly recognized. Good explanations
are believed to improve user trust, loyalty, satisfaction, and
persuasiveness [6]. Experimental evidence has shown that
explanations can improve users' acceptance [
The problem of explaining recommendations is not new.
Several algorithms have been presented in the past and
extensive surveys can be found in [3, 6]. Herlocker et al., [
In this work we present a \black box " explanations
framework based on tags (e.g., as in [
ALGORITHMIC OVERVIEW
Given a recommended item and its tags, we select a small subset of tags that serve best as explanations. These tags are then utilized to automatically generate the explanation sentence. Hence, explanations are computed in two stages: First, we compute scores for the candidate tags to quantify their ability to serve as explanations (\explanatory power"). Then, the tags are ranked and the top tags are chosen to constitute the explanation sentence by applying a prede ned rule-set on tag categories.
The rst stage integrates several modules which constitute scores for either the amount of personalization, quality of the explanation, or quality of the available tags in the context of the recommendation. Our framework allows plugging different modules and tuning their relative contribution with corresponding module weights. Each module is represented as a vector or a matrix and can be of one of the following types: a) Tag modules - vectors fk; b) Item modules -vectors ri; c) Items to tags relationship modules - matrices Pj ; Our frameworks allows for multiple modules per each type above and we denote by k; i; j the indexes of the modules per type.
The scoring algorithm applies an ensemble approach which treats each module as an estimator and computes the weighted average. Let m; n be the number of tags and items, accordingly. Then, sm 1 =
X wiri1 n i !
X wj Pjn m j !
! X wkfkm 1 ; k (1) and s is the resulting scores vector1. Namely, s(l) is the score quantifying the \explanation power" of tag l (see gure 1). The weights wi; wj ; wk are used to tune the importance of 1Note the interchanging use of both the dot product and cross product. each of the modules. For example, by increasing wk we can put more emphasis on the module fk.
Aggregated Item strength ( ) .034 .102 . . . . . . . . . .910 ×
Aggregated Item-Gene correlations ( ) 0.34 … 0.24 .
. . . . . . . . . . . . . . . . . . . . . .
. . .
. 0.91 … -0.1
Aggregated
Gene strength ( ) ⋅
= 0.34 0.12 . . . . 0.91 4.31 3.42 . . . . . . . . . . . . 0.03 tags from di erent categories are chosen to constitute the explanation sentence.
Each item module r1 n weights the items based on their i relevance to the recommended item.
similar to the recommended item are more important than items which are very di erent from the recommended item. A another example for an item module is a module that gives higher weight to items that the user likes more { e.g., Xbox games which are played more frequently should have a higher weight. Information from the di erent item modules is aggregated into an aggregated tag term P i wiri . by the aggregated relationship matrix P
The items to tags relationship modules Pn m are used to j encode the a module P i wiri nity of items to tags. The aggregated items 1 n is projected onto the tags by multiplying j wjPjn m . This results in a tag weights vector which encodes both information from the item
modules and the items to tags relationship modules. Some examples of items to tags modules are: 1) a binary matrix which indicates tags association to items; 2) weighted items to tags strengths which put more emphasis on rare tags.
The tag weights are adjusted by a point-wise multiplication with the aggregated tag modules P k wkfkm 1 . The tags modules score tags based on their ability to serve as a good explanation.
For example, weighting tags based on categories e.g., movie director tags may have more \explanatory power" than plot tags. Additional examples of tag modules are: 1) prior multiplier for speci c tags which serve as better / worst
t for explanations; 2) TF-IDF weights based on item descriptions; 3) a binary or weighted vector signifying the relevance to the recommended item.
In order to construct the explanation sentence we de ne a sentence structure which is composed of 3-4 \slots". Each slot should be matched with a tag according to prede ned tag categories.
We rank the tags according to s and select the highest ranking tags from the appropriate categories to ll in the slots. Note that the lowest ranking tags, those with negative scores but high absolute values, can be used as a negation e.g., \non-dramatic movies". Finally, a rule-set is applied to place the appropriate connective words between the slots according to the tag categories.
We augment each recommended item with an automatically generated personalized explanation sentence (see gure 2).
The explanations are based on tags that describe the recommended item and showed dominance in the user's historical usage patterns. This results in a more meaningful and engaging explanation sentence.
After Based on your history of playing Side view arcade fighting games Based on your history of playing 1st person stealthaction games Wethink you might like Cartoonish 3rd person platformer games the Xbox One recommender system. 4. ltering CSCW '00, 2000. recommendations: Variational bayes matrix factorization with embedded feature selection. In Proc. collaborative