解析可解释推荐系统：历史、现状与未来趋势

需积分: 24 177 浏览量更新于2024-07-15 收藏 16.31MB PDF 举报

可解释推荐系统（Explainable Recommendation）是一种致力于开发既能提供高质量推荐，又能给出直观解释的智能技术。该领域的研究始于对传统推荐算法透明度不足问题的关注，随着用户对于推荐背后逻辑的理解需求增加，解释性成为现代推荐系统设计中的关键要素。文章《ExplainableRecommendation:A Survey and New Perspectives》详细回顾了这一领域的发展历程，从早期的历史背景到2020年2月的最新进展。历史上，可解释推荐系统的重点在于解决“为什么”问题，即解释推荐算法为何选择特定项目。这主要通过两种方式实现：一是后知后觉的解释（post-hoc），即在推荐结果生成后，提供针对推荐决策的解释；二是直接来自解释性模型（也称为可解释或透明模型），这类模型本身就能生成易于理解的决策依据。当前的研究现状聚焦于模型方法的发展，特别是模型基础方法的大量提出和应用。这些方法包括但不限于基于规则的解释、基于案例的学习、特征重要性分析、局部可解释性模型（如树模型和线性模型）、以及最近的深度学习模型的可解释性增强策略，如注意力机制和可视化技术。评价可解释推荐系统的标准通常涉及以下几个方面：透明度（Transparency）、说服力（Persuasiveness）、有效性（Effectiveness）、可信度（Trustworthiness）和满意度（Satisfaction）。透明度评估推荐系统的解释是否清晰易懂，说服力关注用户是否接受并信任推荐，有效性则衡量推荐结果的质量。同时，解释性还有助于系统调试，帮助开发者更好地识别和修复潜在问题。未来发展趋势方面，可解释推荐系统将继续融合深度学习与解释性技术，探索更深层次的因果关系理解和用户个性化需求的满足。此外，随着数据隐私和伦理问题日益突出，如何在保护用户隐私的同时提供有效的解释也将是重要的研究方向。可解释推荐系统作为提升用户理解和信任、推动推荐系统优化的重要手段，正逐步成为信息技术领域的一个研究热点。随着技术的不断进步和应用场景的扩展，它将在未来的推荐服务中扮演越来越重要的角色。

16 Information Source for Explanations

Figure 2.2:

An example of explanations based on relevant users. (a) A histogram

of the neighbors’ ratings are displayed as an explanation for the recommended

item, where the positive and negative ratings are correspondingly clustered, and

the neutral ratings are displayed separately. Based on this explanation, a user can

easily understand that the item is recommended because his/her neighbors made

high ratings on the item. (b) An explanation for the movie “The Sixth Sense”, where

each bar represents the rating of a neighbor, and the

-axis represents a neighbor’s

similarity to the user. With this explanation, it would be easy to understand how a

user’s most similar neighbors rated the target movie (Herlocker et al., 2000).

Figure 2.3:

A comparison between relevant-item explanation (left) and relevant-user

explanation (right) (Abdollahi and Nasraoui, 2017).

for the recommended item (i.e., the camera lens), a relevant-user expla-

nation tells Bob that similar users William and Fred also bought this

item, while a relevant-item explanation persuades Bob to buy the lens

by showing the camera he already bought before.

To study how explanations help in recommender systems, Tintarev

(2007) developed a prototype system to study the eﬀect of diﬀerent

types of explanations, especially the relevant-user and relevant-item

explanations. In particular, the author proposed seven beneﬁts of provid-

ing recommendation explanations, including transparency, scrutability,

trustworthiness, eﬀectiveness, persuasiveness, eﬃciency, and satisfaction.

The version of record is available at: http://dx.doi.org/10.1561/1500000066

2.2. Feature-based Explanation 17

Based on their user study, the author showed that providing appropriate

explanations can indeed beneﬁt the recommender system over these

seven perspectives.

Usually, relevant-item explanations are more intuitive for users to

understand because users are familiar with the items they interacted

before. As a result, these items can serve as credible explanations for

users. Relevant-user explanations, however, could be less convincing

because the target user may know nothing about other “similar” users

at all, which may decrease the trustworthiness of the explanations

(Herlocker et al., 2000). Besides, disclosing other users’ information

may also cause privacy problems in commercial systems. This problem

drives relevant-user explanation into a new direction, which leverages

social friend information to provide social explanations (Ren et al., 2017;

Tsai and Brusilovsky, 2018). For example, we can show a user with her

friends’ public interest as explanations for our social recommendations.

In the following, we will review this research direction in the social

explanation section (Section 2.6).

2.2 Feature-based Explanation

The feature-based explanation is closely related to content-based rec-

ommendation methods. In content-based recommendation, the system

provides recommendations by matching the user proﬁle with the content

features of candidate items (Cramer et al., 2008; Ferwerda et al., 2012;

Pazzani and Billsus, 2007). Content-based recommendations are usually

intuitive to explain based on the features.

Depending on the application scenario, content-based recommenda-

tions can be generated from diﬀerent item features. For example, movie

recommendations can be generated based on movie genres, actors, or

directors; while book recommendations can be provided based on book

types, price, or authors. A conventional paradigm for feature-based

explanation is to show users with the features that match the user’s

proﬁle.

Vig et al. (2009) adopted movie tags as features to generate recom-

mendations and explanations, as shown in Figure 2.4. To explain the

recommended movie, the system displays the movie features and tells

The version of record is available at: http://dx.doi.org/10.1561/1500000066

2.3. Opinion-based Explanation 19

an item is recommended and why others are not. As shown in Figure 2.5,

a recommendation is explained in that most of its aspects satisfy the

preference of the target user.

User demographic information describes the content features of users,

and the demographic features can also be used to generate feature-based

explanations. Demographic-based recommendation (Pazzani, 1999) is

one of the earliest approaches to personalized recommendation systems.

Recently, researchers have also integrated demographic methods into

social media to provide product recommendations in social environments

(Zhao et al., 2014, 2016).

The demographic-based approach makes recommendations based on

user demographic features such as age, gender, and residence location.

Intuitively, an item recommended based on demographic information

can be explained by the demographic feature(s) that triggered the

recommendation, e.g., by telling the user that “80% of customers in

your age bought this product”. Zhao et al. (2014) represented products

and users in the same demographic feature space, and used the weights

of the features learned by a ranking function to explain the results;

Zhao et al. (2016) further explored demographic information in social

media environment for product recommendation with feature-based

explanations.

2.3 Opinion-based Explanation

More and more user-generated contents have been accumulating on the

Web, such as e-commerce reviews and social media posts, which help

users to express their opinion on certain items or aspects. Researchers

have shown that such information is quite beneﬁcial in user proﬁling

and recommendation (McAuley and Leskovec, 2013; Zheng et al., 2017).

Besides, it helps to generate ﬁner-grained and more reliable explanations,

which beneﬁt users to make more informed decisions (Li et al., 2017;

Zhang et al., 2014a). With this motivation, many models have been

proposed to explain recommendations based on user-generated texts.

Methods in this direction can be broadly classiﬁed into aspect-level

and sentence-level approaches, according to how the explanations are

displayed. See Figure 2.1 for example, where aspect-level models present

The version of record is available at: http://dx.doi.org/10.1561/1500000066

20 Information Source for Explanations

item aspects (such as color, quality) and their scores as explanations,

while the sentence-level models directly present an explanation sentence

to users about why the camera lens is recommended. We will focus on

aspect-level explanation in this subsection, while sentence-level explana-

tions will be introduced in the following subsection together with other

natural language generation-based explanation models.

The aspect-level explanation is similar to feature-based explanation,

except that aspects are usually not directly available in an item or user

proﬁle. Instead, they are extracted or learned as part of the recommen-

dation model from – e.g., the reviews – and the aspects can be paired

up with consumer opinions to express a clear sentiment on the aspect.

Researchers in the sentiment analysis community have explored

both data mining and machine learning techniques for aspect-level

sentiment analysis, which aims to extract aspect-sentiment pairs from

text. For example, Hu and Liu (2004) proposed a frequent feature

mining approach to aspect-level sentiment analysis, and Lu et al. (2011)

proposed an optimization approach to construct context-aware sentiment

lexicons automatically. A comprehensive review on sentiment analysis

and opinion mining is summarized in Liu (2012). Based on these research

eﬀorts, Zhang et al. (2014b) developed a phrase-level sentiment analysis

toolkit named Sentires

to extract “aspect–opinion–sentiment” triplets

from reviews of a product domain. For example, given large-scale user

reviews about mobile phones, the toolkit can extract triplets such

as “noise–high–negative”, “screen–clear–positive”, and “battery_life–

long–positive”. The toolkit can also detect the contextual sentiment of

the opinion words under diﬀerent aspect words. For example, though

“noise” paired with “high” usually represents a negative sentiment, when

“quality” is paired with “high”, however, it instead shows a positive

sentiment. Based on the dictionary of aspect–opinion–sentiment triplets

constructed by the program, it can further detect which triplets are

contained in a review sentence.

Based on this toolkit, researchers developed diﬀerent models for ex-

plainable recommendation. For example, Zhang et al. (2014a) proposed

an explicit factor model for explainable recommendation, which presents

http://yongfeng.me/software/

The version of record is available at: http://dx.doi.org/10.1561/1500000066

剩余100页未读，继续阅读

qq_28147259

粉丝: 2
资源: 13

解析可解释推荐系统：历史、现状与未来趋势

推荐系统调研报告及综述（张永锋）

改善偏好预测：可调可解释的推荐系统（Amber Roberts）

陈旭__Dynamic Explainable Recommendation based on Neural Attentive

Explainable Deep Learning Methods in Medical Diagnosis：A Survey.

1-2+Knowledge+Graph+Based+Explainable+Recommendation.pdf

Jointly Learning Explainable Rules for Recommendation with Knowledge Graph.pdf

Explainable Reasoning over Knowledge Graphs for Recommendation.pdf

Explainable_AI：可解释的AI笔记本和项目的集合

explainable_cdss:可解释的CDSS可再现性

基于偏差网络的可解释深部少炮异常检测_Explainable Deep Few-shot Anomaly Detection w

最新资源