The document describes a method for generating natural language justifications for recommender systems using text summarization and sentiment analysis techniques. It discusses prior approaches using descriptive properties or review-based features, and proposes a new approach that exploits automatic text summarization of user reviews. The proposed method involves extracting aspects from reviews, ranking aspects based on frequency, sentiment, and importance, and generating a summary justification using a centroid-based text summarization algorithm on filtered sentences from reviews. The goal is to provide a higher-quality justification by summarizing relevant information from multiple reviews.
Hyperautomation and AI/ML: A Strategy for Digital Transformation Success.pdf
Natural Language Justifications for Recommender Systems Exploiting Text Summarization and Sentiment Analysis
1. @cataldomusto cataldo.musto@uniba.it
Natural Language Justifications for
Recommender Systems Exploiting Text
Summarization and Sentiment Analysis
CATALDO MUSTO, GAETANO ROSSIELLO, MARCO DE GEMMIS, PASQUALE LOPS AND GIOVANNI SEMERARO
UNIVERSITÀ DEGLI STUDI DI BARI ALDO MORO - ITALY
2. Recommender
Systems
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019 2
3. The Explanation Problem
Recommendation
I suggest you…
3Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
4. The Explanation Problem
Recommendation
4Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
5. Early approaches[*]: descriptive properties
Recommendation
I suggest you The Ring because you
often like movies with Naomi Watts
as 21 grams and Mulholland Drive.
Furthermore, you like films about
ghosts such as The Sixth Sense.
5Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
[*] Cataldo Musto, Fedelucio Narducci, Pasquale Lops, Marco de Gemmis, Giovanni
Semeraro: ExpLOD: A Framework for Explaining Recommendations based on
the Linked Open Data Cloud. RecSys 2016: 151-154
6. More recently[*]: review-based features
I recommend you The Ring because
people who liked the movie think that
it delivers some bone-chilling terror.
Moreover, people liked The Ring
since the casting is pretty good.
6Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
[*] Cataldo Musto, Pasquale Lops, Marco de Gemmis, Giovanni Semeraro:
Justifying Recommendations through Aspect-based Sentiment Analysis of
Users Reviews. UMAP 2019: 4-12
7. More recently: review-based features
To identify relevant and distinguishing
characteristics of the recommended
item by mining users’ reviews
7Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
8. Why should we use reviews?
Intense thriller
Pretty good casting
Well-plotted investigation
Impressive horror
......
To identify relevant and distinguishing
characteristics of the recommended
item by mining users’ reviews
8Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
9. Why should we use reviews?
To identify relevant and distinguishing
characteristics of the recommended
item by mining users’ reviews
9Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
I recommend you The Ring because
people who liked the movie think that it
delivers some bone-chilling terror.
Moreover, people liked The Ring since the
casting is pretty good.
10. Cataldo Musto. Advances in Content-based Recommender Systems – Explanation Strategies.
ACM Summer School on Recommender Systems 2019, Gothenburg (Sweden), September 13 2019
Our first methodology has
two main weaknesses
• Very naïve strategy for detecting more
relevant aspects
• Very static template-based to
generate natural language
explanations
10
Why do we need another approach?
11. Review-based Explanations
exploiting Automatic Text
Summarization
In this talk
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni
Semeraro: Combining Text Summarization and Aspect-based Sentiment
Analysis of Users’ Reviews to Justify Recommendations.
ACM RecSys 2019, pp. 383-387
11Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
12. To exploit automatic text summarization
techniques to build an higher-quality justifications.
Intuition
12Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
13. To exploit automatic text summarization
techniques to build an higher-quality justifications.
We conceive our justification as a summary of the
information conveyed by all the available reviews.
Intuition
13Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
14. Workflow
14Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
15. Workflow
15Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
16. Aspect Extraction
Goal: to identify the aspects that are
discussed when people talk about the item
16
reviews aspects
Input: reviews of the item i R = {ri1, ri2 … rin}
Output: aspects A = {ai1, ai2 … aik}
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
17. Aspect Extraction
Statistical approach based on the Kullback-Leibler
(KL) Divergence
Measures the difference between the distribution of a term
in a generic corpus (e.g. BNC) and its distribution in a domain corpus
(e.g. movie reviews)
17Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
18. Aspect Extraction
Measures the difference between the distribution of a term
in a generic corpus (e.g. BNC) and its distribution in a domain corpus
(e.g. movie reviews)
Insight: the higher the divergence, the higher the
importance of the term in the domain
t = term
ca = corpus A
cb = corpus B
Statistical approach based on the Kullback-Leibler
(KL) Divergence
18Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
19. Aspect Extraction
Measures the difference between the distribution of a term
in a generic corpus (e.g. BNC) and its distribution in a domain corpus
(e.g. movie reviews)
Insight: the higher the divergence, the higher the
importance of the term in the domain
KL(cast, BNC, movie-reviews) >> 0
KL(actor, BNC, movie-reviews) > 0
KL(city, BNC, movie-reviews) ~ 0
KL(woman, BNC, movie-reviews) ~ 0
We label as ‘aspects’ the
nouns whose
KL-divergence is higher
than zero
Statistical approach based on the Kullback-Leibler
(KL) Divergence
19Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
20. Aspect Extraction
Measures the difference between the distribution of a term
in a generic corpus (e.g. BNC) and its distribution in a domain corpus
(e.g. movie reviews)
Insight: the higher the divergence, the higher the
importance of the term in the domain
KL(cast, BNC, movie-reviews) >> 0 YES
KL(actor, BNC, movie-reviews) > 0 YES
KL(city, BNC, movie-reviews) ~ 0 NO
KL(woman, BNC, movie-reviews) ~ 0 NO
Statistical approach based on the Kullback-Leibler
(KL) Divergence
20Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
21. Aspect Ranking
21Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
22. Aspect Ranking
aspects top-k aspects
Input: aspects A = {ai1, ai2 … aim}
Output: top-k aspects A = {ai1, ai2 … aik}
22Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
23. Aspect Ranking
Goal: to identify distinguishing aspects that are discussed
with a positive sentiment when people talk about the item
23Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
24. Aspect Ranking
Goal: to identify distinguishing aspects that are discussed
with a positive sentiment when people talk about the item
24
How many times aspect ‘a’ appears in the
reviews of item ‘i’ (frequency of the aspect)
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
25. Aspect Ranking
Goal: to identify distinguishing aspects that are discussed
with a positive sentiment when people talk about the item
25
How positive is the opinion of the users
when they talk about aspect ‘a’ (opinion
towards the aspect)
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
26. Aspect Ranking
Goal: to identify distinguishing aspects that are discussed
with a positive sentiment when people talk about the item
26
How distinguishing is
the aspect ‘a’
(KL-divergence score)
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
27. Aspect Ranking
Goal: to identify distinguishing aspects that are discussed
with a positive sentiment when people talk about the item
27
Intuition: our formula gives an higher score to the aspects that are
frequently mentioned in the reviews with a positive sentiment.
Moreover, it also rewards less popular aspects (higher KL-DIV).
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
28. Generation
28Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
29. Generation
Intuition: we conceive our justification as a summary of the
information conveyed by all the available reviews
29Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
30. Generation
Intuition: we conceive our justification as a summary of the
information conveyed by all the available reviews
Approach: we exploited a centroid-based method for automatic text
summarization. Very good performance in multi-document
summarization scenarios.
Assumption: each review can be considered as ‘document’ thus the
corpus of the reviews can be used to feed the algorithm
30Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
31. Generation
Generation process is in turn split into two steps
• Sentence Filtering
• Text Summarization
Sentence Filtering is used to feed the summarization algorithm
with compliant sentences. We selected sentences that matched
the following criterions:
31Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
32. Generation
Generation process is in turn split into two steps
• Sentence Filtering
• Text Summarization
Sentence Filtering is used to feed the summarization algorithm
with compliant sentences. We selected sentences that matched
the following criterions:
• The sentence contains a main aspect
• The sentence is longer than 5 tokens
• The sentence expresses a positive sentiment
• The sentence does not contain first-person personal or possessive pronouns
32Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
33. Generation
Text Summarization Algorithm
Input: item i, sentences s1…sn, word limit k
Output: summary for item i consisting of k words
1. Build a vector space representation for each sentence
2. Merge all the sentences in a pseudo-document that represents the
item (centroid vector)
3. Until the word limit is not reached
3.1 Go through the sentences and calculate cosine similarity
3.2 Pick the one with the highest cosine similarity to the
centroid (and not that similar to the sentences previously
picked)
3.3 Add it to the summary
33Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
34. Generation
Text Summarization Algorithm
Input: item i, sentences s1…sn, word limit k
Output: summary for item i consisting of k words
1. Build a vector space representation for each sentence
2. Merge all the sentences in a pseudo-document that represents the
item (centroid vector)
3. Until the word limit is not reached
3.1 Go through the sentences and calculate cosine similarity
3.2 Pick the one with the highest cosine similarity to the
centroid (and not that similar to the sentences previously
picked)
3.3 Add it to the summary
34Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
35. Generation
Text Summarization Algorithm
Input: item i, sentences s1…sn, word limit k
Output: summary for item i consisting of k words
1. Build a vector space representation for each sentence
2. Merge all the sentences in a pseudo-document that represents the
item (centroid vector)
3. Until the word limit is not reached
3.1 Go through the sentences and calculate cosine similarity
3.2 Pick the one with the highest cosine similarity to the
centroid (and not that similar to the sentences previously
picked)
3.3 Add it to the summary
35Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
36. Generation
Text Summarization Algorithm
Input: item i, sentences s1…sn, word limit k
Output: summary for item i consisting of k words
1. Build a vector space representation for each sentence
2. Merge all the sentences in a pseudo-document that represents the
item (centroid vector)
3. Until the word limit is not reached
3.1 Go through the sentences and calculate cosine similarity
3.2 Pick the one with the highest cosine similarity to the
centroid (and not that similar to the sentences previously
picked)
3.3 Add it to the summary
36Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
37. Generation
Text Summarization Algorithm
Input: item i, sentences s1…sn, word limit k
Output: summary for item i consisting of k words
1. Build a vector space representation for each sentence
2. Merge all the sentences in a pseudo-document that represents the
item (centroid vector)
3. Until the word limit is not reached
3.1 Go through the sentences and calculate cosine similarity
3.2 Pick the one with the highest cosine similarity to the
centroid (and not that similar to the sentences previously
picked)
3.3 Add it to the summary
37Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
38. Generation – Final Output
“If you like or love the blood and gore kinds of films,
this movie will certainly disappoint you as the focus is
on character, story, mood and unique special effects.
The Ring is a story about supernatural evil therefore,
it is a horror film, done very much in the style of the
psychological thriller.”
Legenda
Red: aspects (k=4)
Black: compliant excerpts
38Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
39. Experimental Evaluation
Research Question 1
How effective are the justifications generated through the pipeline, on varying of different
combinations of the parameters?
Research Question 2
How does our justifications perform with respect to a simple review-based explanation?
Experimental Design
User Study with a Web Application
141 subjects
Movie Domain. 300 movies. ~150k reviews.
Metrics: Transparency, Engagement, Persuasion, Trust, Effectiveness [^]
Parameters: Justification Length (Short=50 words, Long=100) and #Aspects (10 and 30).
Between-subjects for Research Question 1, Within-subjects for Research Question 2
[^] Tintarev, N., & Masthoff, J. Designing and evaluating
explanations for recommender systems. In Recommender
systems handbook. pp. 479-510. Springer, Boston, MA. 2011
39Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
40. Experimental Protocol (Research Question 1)
1. Gathering movie preferences
Users rated their favourite movies
40Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
41. Experimental Protocol (Research Question 1)
2. Recommendation is obtained
Personalized PageRank as algorithm
1. Gathering movie preferences
Users rated their favourite movies
41Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
42. Experimental Protocol (Research Question 1)
2. Recommendation is obtained
Personalized PageRank as algorithm
3. Explanation is generated
Random Configuration (users not aware)
1. Gathering movie preferences
Users rated their favourite movies
42Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
43. Experimental Protocol (Research Question 1)
1. Gathering movie preferences
Users rated their favourite movies
2. Recommendation is obtained
Personalized PageRank as algorithm
3. Explanation is generated
Random Configuration (users not aware)
4. Metrics are calculated
Through Questionnaires
43Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
44. Results (Research Question 1)
MOVIES Transparency Persuasion Engagement Trust Effectiveness
Top-10 Short 2.83 3.06 3.06 2.83 0.89
Top-30 Long 3.16 3.06 2.69 3.19 0.94
Top-10 Short 3.95 3.64 3.37 3.55 0.55
Top-30 Long 3.24 3.18 3.12 3.22 0.38
Finding 1
Long justifications better
than short justifications,
on average
Finding 2
Top-10 aspect provide
better explanations than
Top-30 aspects
Finding 3
Long explanations based
on Top-10 aspects lead to
the best results
44Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
45. Experimental Protocol (Research Question 2)
Review-based
Explanation +
Summarization
Review-based
Explanation
45Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
46. Results (Research Question 2)
MOVIES
Review+
Summary
Review-
based
Indiffer.
Transparency 54.5% 40.9% 4.6%
Persuasion 77.3% 13.6% 9.1%
Engagement 63.6% 27.3% 9.1%
Trust 68.2% 4.5% 27.3%
Outcome: automatic Text Summarization provides
users with the best explanation. Confirmed for all the
metrics and both the domains
46Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
47. Recap
“If you like or love the blood and
gore kinds of films, this movie
will certainly disappoint you as
the focus is on character, story,
mood and unique special
effects. The Ring is a story about
supernatural evil therefore, it is a
horror film, done very much in
the style of the psychological
thriller.”
Review-based
explanation
using
automatic text
summarization
47Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019
48. Take-home Messages
48
1.
2.
Textual features extracted from reviews can be used to generate a
natural language justifications supporting a recommendation
Cataldo Musto, Gaetano Rossiello, Marco de Gemmis, Pasquale Lops, Giovanni Semeraro. Natural Language Justifications for
Recommender Systems Exploiting Text Summarization and Sentiment Analysis. AI*IA 2019, Rende, November 22, 2019 48
Text summarization algorithms are suitable to build a less ‘static’
justifications based on review excerpts
3. Shorter Justifications based on Top-10 aspects obtained the best
results