A General Aspect-Term-Extraction Model for Multi-Criteria
Recommendations
Paolo Pastore
1
, Andrea Iovine
2
, Fedelucio Narducci
1
and Giovanni Semeraro
2
1
Polytechnic University of Bari, Italy
2
Dept. of Computer Science University of Bari, Italy
Abstract
In recent years, increasingly large quantities of user reviews have been made available by several e-commerce platforms.
This content is very useful for recommender systems (RSs), since it refects the users’ opinion of the items regarding several
aspects. In fact, they are especially valuable for RSs that are able to exploit multi-faceted user ratings. However, extracting
aspect-based ratings from unstructured text is not a trivial task. Deep Learning models for aspect extraction have proven
to be efective, but they need to be trained on large quantities of domain-specifc data, which are not always available. In
this paper, we explore the possibility of transferring knowledge across domains for automatically extracting aspects from
user reviews, and its implications in terms of recommendation accuracy. We performed diferent experiments with several
Deep Learning-based Aspect Term Extraction (ATE) techniques and Multi-Criteria recommendation algorithms. Results
show that our framework is able to improve recommendation accuracy compared to several baselines based on single-criteria
recommendation, despite the fact that no labeled data in the target domain was used when training the ATE model.
Keywords
multi-criteria recommendation, deep learning, aspect term extraction, domain adaptation, transfer learning
1. Introduction
Nowadays, many Web platforms and e-commerce web-
sites allow customers to express their opinions by pro-
viding reviews on items, services, or media. Such user-
generated content is extremely valuable for recommen-
dation, since it refects the user’s perception of a spe-
cifc item and of specifc features of that item listing
its strengths and weaknesses, the most important fea-
tures, and the tasks for which it is more (or less) suitable.
Extracting this information and exploiting it to enrich
user profles and item descriptions can give enormous
advantages to Recommender Systems (RSs). Given the
considerable importance of reviews in the recommen-
dation process, many works in the literature proposed
the idea of integrating them into RSs, as a way to im-
prove their accuracy. Specifcally, text reviews can be a
solution to the rating sparsity problem ofen encountered
by RSs based on Collaborative Filtering (CF), and can
be used to capture a much more fne-grained model of
the customer’s preferences [1]. Accordingly, instead of
modeling the user’s profle as a set of (item, rating) pairs,
it might be represented as a set of (item, aspect, rating)
triples. Of course, the problem with this approach is that
3rd Edition of Knowledge-aware and Conversational Recommender
Systems (KaRS) & 5th Edition of Recommendation in Complex
Environments (ComplexRec) Joint Workshop @ RecSys 2021,
September 27–1 October 2021, Amsterdam, Netherlands
E paolo.pastore1@poliba.it (P. Pastore); andrea.iovine@uniba.it
(A. Iovine); fedelucio.narducci@poliba.it (F. Narducci);
giovanni.semeraro@uniba.it (G. Semeraro)
© 2021 Copyright for this paper by its authors. Use permitted under Creative
Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings CEUR Workshop Proceedings (CEUR-WS.org)
both aspects and ratings must be extracted automatically
from unstructured text. This task is usually referred to
as Aspect-Based Sentiment Analysis (ABSA). ABSA is not
a trivial task, because there is no stable defnition of ”as-
pect”, due to its intrinsic subjectivity. Also, the same
aspect can appear in many diferent forms inside user
reviews. For instance, a reviewer could use ”service”,
”staf” or ”waiter” for referring to the ”service” category.
For this reason, we distinguish between the aspect itself
and its representation forms in the reviews, also called
aspect terms. Furthermore, the aspects used in a domain
are completely diferent to those in other domains: for
restaurants, users will mention features such as the food
or the quality of the service, when talking about smart-
phones, they will instead refer to other aspects such as
the screen or the camera. In recent years, many models
for automatically extracting aspects from text based on
Deep Learning models have been proposed. However,
these techniques need to be trained on domain-specifc
labeled datasets that are not always available.
In this paper, we investigate the application of domain
adaptation strategies for aspect-based recommendation.
The aim is to evaluate the efectiveness of modern Deep
Learning-based Aspect Term Extraction (ATE) models
when no annotated data is available for the target do-
main. For this purpose, we developed an aspect-based
recommendation framework that includes an ATE mod-
ule, an Aspect Clustering module, a Sentiment Analysis
(SA) module, and a Multi-Criteria Recommender Sys-
tem. We performed an experimental study to compare
several ATE models both in a single domain scenario
and in a domain adaptation setting. We then chose the