Experiencer-Specific Emotion and Appraisal Prediction Maximilian Wegge Enrica Troiano Laura Oberländer and Roman Klinger Institut für Maschinelle Sprachverarbeitung University of Stuttgart

2025-05-06 0 0 353.18KB 8 页 10玖币

侵权投诉

Experiencer-Speciﬁc Emotion and Appraisal Prediction

Maximilian Wegge, Enrica Troiano, Laura Oberländer and Roman Klinger

Institut für Maschinelle Sprachverarbeitung, University of Stuttgart

{firstname.lastname}@ims.uni-stuttgart.de

Abstract

Emotion classiﬁcation in NLP assigns emotions

to texts, such as sentences or paragraphs. With

texts like “I felt guilty when he cried”, focusing

on the sentence level disregards the standpoint

of each participant in the situation: the writer

(“I”) and the other entity (“he”) could in fact

have different affective states. The emotions

of different entities have been considered only

partially in emotion semantic role labeling, a

task that relates semantic roles to emotion cue

words. Proposing a related task, we narrow

the focus on the experiencers of events, and as-

sign an emotion (if any holds) to each of them.

To this end, we represent each emotion both

categorically and with appraisal variables, as a

psychological access to explaining why a per-

son develops a particular emotion. On an event

description corpus, our experiencer-aware mod-

els of emotions and appraisals outperform the

experiencer-agnostic baselines, showing that

disregarding event participants is an oversim-

pliﬁcation for the emotion detection task.

1 Introduction

Computational emotion analysis from text includes

various subtasks, with the most prominent one be-

ing emotion classiﬁcation or regression. Its goal is

to assign an emotion representation to textual units,

and the way this is done typically depends on the

domain of the data, the practical application of the

task, and the psychological theories of reference:

emotions can be modelled as discrete labels, in

line with theories of basic emotions (Ekman,1992;

Plutchik,2001), as valence–arousal value pairs that

deﬁne an affect vector space where to situate emo-

tion concepts (illustrated, e.g., by Posner et al.,

2005), or as appraisal spaces that correspond to the

cognitive evaluative dimensions underlying emo-

tions

(Scherer,2005;Smith and Ellsworth,1985).

They are similar to a valence–arousal space, but the di-

mensions correspond to evaluations of events (i.e., appraisals)

that underlie a certain emotion.

Irrespective of the adopted representations, most

work in the ﬁeld detects emotions from a single

perspective – either to recover the emotion that the

writer of a text likely expressed (e.g., with respect

to emotion categories and intensities (Mohammad

et al.,2018), and cognitive categories (Hofmann

et al.,2020)), or to predict the emotion that the

text elicits in the readers (e.g., using news arti-

cles, Strapparava and Mihalcea,2007;Bostan et al.,

2020). Only a few approaches combine or compare

the reader’s with the writer’s perspective (Buechel

and Hahn,2017, i.a.). However, none of them looks

at the perspectives of the participants in events

(both mentioned or implicit) as described by a text.

Focusing on such perspectives separately is es-

sential to develop an all-round account of the af-

fective implications that events have. It would em-

phasize how the facts depicted in text are amenable

to different “emotion narratives”, by pushing one

or the other perspective in the foreground. For

instance, a possible interpretation for the sen-

tence “As the waiter yelled at her, the expression

on my mother’s face made all the staff look re-

pulsed”, could be: “my mother”

→

sadness, “the

waiter”

→

anger, and “the staff”

→

disgust. There,

one entity is responsible for an event (screaming),

one is inﬂuenced by it, and the third is affected

by the emotion emerging in the other (the facial

expression, which can be seen as an event in itself).

Our goal is close to emotion role labeling, a

special case of semantic role labeling (SRL) (Mo-

hammad et al.,2018;Kim and Klinger,2018). SRL

addresses the question “Who did What to Whom,

Where, When, and How?” (Gildea and Jurafsky,

2000), emotion SRL asks “Who feels what, why,

and towards whom?” (Kim and Klinger,2018),

mainly to detect causes of emotion-eliciting events

(Ghazi et al.,2015) for certain entities. Here,

we tackle a variation of this question, namely,

“Who feels what and under which circumstances?”.

The circumstances refer to the explanation pro-

arXiv:2210.12078v2 [cs.CL] 26 May 2023

vided by appraisal interpretations, another novelty

that we contribute to the emotion SRL panorama.

Appraisal-based emotion representations capture

entity-speciﬁc aspects that lead to an emotion, as

they describe the subjective qualities that an indi-

vidual sees in events.

We propose a method for experiencer-speciﬁc

emotion and appraisal analysis that bridges emo-

tion classiﬁcation and semantic role labeling.

Given texts that describe events and that include an-

notations for all participants, we assign an emotion

and an appraisal vector to each potential emoter.

Our proposal is computationally simpler than cre-

ating a full graph of relations between causes and

entities, as is normally done in (emotion) SRL. Yet,

its ﬁne-grained focus on event participants is bene-

ﬁcial over traditional classiﬁcation- and regression-

based approaches: by predicting an emotion and

scoring multiple appraisals for each entity, our

model strongly outperforms text-level baselines.

Thus, the results demonstrate that assigning one

emotion to the entire instance, or multiple emo-

tions without considering for whom they hold, is a

simpliﬁcation of the emotional import of the text.

2 Related Work

In natural language processing, emotions are usu-

ally represented as discrete names following the-

ories of basic emotions (Ekman,1992;Plutchik,

2001), or as values of valence and arousal (Rus-

sell and Mehrabian,1977). Computational models

based on such representations have been applied

to many text sources, including Reddit comments

(Demszky et al.,2020) and tales (Alm et al.,2005),

but also to resources created as part of psychologi-

cal research. An example is the ISEAR corpus. It

consists of short reports collected in lab (Scherer

and Wallbott,1997), instructing participants to de-

scribe events that caused in them a certain emo-

tion. A similar collection practice was adopted

by Troiano et al. (2019). In their enISEAR, crowd-

workers completed sentences like “I felt [EMOTION

NAME] when . . . ” for seven emotion names.

The emotions of entities are considered in emo-

tion SRL, whose goals comprise the recognition of

emotion cue words, emotion experiencers/emoters

and descriptions of emotion causes and targets (Mo-

hammad et al.,2018;Bostan et al.,2020;Kim

and Klinger,2018;Campagnano et al.,2022, i.a.).

Yet, most work focused on detecting causes (i.e.,

emotion-triggering events), and less on other se-

Emotion Class # inst. # exp.

anger 259 336

disgust 73 87

fear 173 220

joy, pride, contentment 181 265

no emotion 223 269

other, anticipation, hope,

surprise, trust 102 117

sadness, disappointment,

frustration 320 423

shame, guilt 282 325

total 720 1329

Table 1: Number of instances and experiencer spans

annotated for each emotion. Non-bold emotion names

are concepts in the x-enVENT data that we merge with

bold emotion names in our experiments.

mantic roles (Russo et al.,2011;Chen et al.,2018,

2010;Cheng et al.,2017, i.a.).

The gap between entity-speciﬁc emotion anal-

ysis and emotion SRL was partially ﬁlled in by

Troiano et al. (2022). They aimed at better under-

standing the readers’ attempts to interpret the expe-

rience of the texts’ authors. They post-annotated

instances from enISEAR with emotions and 22 ap-

praisal concepts, both for the writer and all other

event participants mentioned in the text. The ap-

praisal variables include evaluations of events, as

they were likely conducted by the event experi-

encers, including if authors felt responsible, if they

needed to pay attention to the environment, whether

they found themselves in control of the situation,

and its pleasantness (see Table 1 in their paper for

explanations of the variables). However, their work

was limited to corpus creation and analysis, and did

not provide any modeling of appraisals or emotions

in an emotion experiencer-speciﬁc manner. There-

fore, it is not clear whether a simplifying assump-

tion that all entities experience the same emotion

or an actual entity-speciﬁc model performs practi-

cally better. We address this concern and show that

experiencer-speciﬁc modeling is beneﬁcial.

Finally, our work is related to structured sen-

timent analysis (Barnes et al.,2021), in which

opinion targets, their polarity, but also an opinion-

holding (or expressing) entity is to be detected.

Most studies focused on sentiment targets and as-

pects (Brauwers and Frasincar,2021), but there are

also some that aim at detecting the opinion holder

(Kim and Hovy,2006;Wiegand and Klakow,2011;

Seki,2007;Wiegand and Klakow,2012, i.a.).

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Experiencer-SpecificEmotionandAppraisalPredictionMaximilianWegge,EnricaTroiano,LauraOberländerandRomanKlingerInstitutfürMaschinelleSprachverarbeitung,UniversityofStuttgart{firstname.lastname}@ims.uni-stuttgart.deAbstractEmotionclassificationinNLPassignsemotionstotexts,suchassentencesorparagraphs.Wit...

展开>> 收起<<

Experiencer-Specific Emotion and Appraisal Prediction Maximilian Wegge Enrica Troiano Laura Oberländer and Roman Klinger Institut für Maschinelle Sprachverarbeitung University of Stuttgart.pdf

共8页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Experiencer-Specific Emotion and Appraisal Prediction Maximilian Wegge Enrica Troiano Laura Oberländer and Roman Klinger Institut für Maschinelle Sprachverarbeitung University of Stuttgart

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: