We extract a large-scale stance detection dataset from comments written by candidates of elections in Switzerland. The dataset consists of German, French and Italian text, allowing for a cross-lingual evaluation of stance detection. It contains 67 000 comments on more than 150 political issues (targets). Unlike stance detection models that have specific target issues, we use the dataset to train a single model on all the issues. To make learning across targets possible, we prepend to each instance a natural question that represents the target (e.g. “Do you support X?”). Baseline results from multilingual BERT show that zero-shot crosslingual and cross-target transfer of stance detection is moderately successful with this approach.
In recent years many datasets have been
created for the task of automated stance detection,
advancing natural language understanding
systems for political science, opinion research and
other application areas. Typically, such
benchmarks
First of all, it is questionable how well
current stance detection methods perform in a
crosslingual setting, as the multilingual datasets
available today are relatively small, and specific to a
single target (Taulé et al., 2017, 2018).
Furthermore, specific models tend to be developed for
each single target or pair of targets
In this paper we propose a much larger dataset that combines multilinguality and a multitude of topics and targets. X-stance comprises more than 150 questions about Swiss politics and more than 67k answers given by candidates running for political office in Switzerland. Questions are available in four languages: English, Swiss Standard German, French, and Italian. The language of a comment depends on the candidate’s region of origin.
We have extracted the data from the voting advice application Smartvote. Candidates respond to questions mainly in categorical form (yes / rather yes / rather no / no). They can also submit a freetext comment to justify or explain their categorical answer. An example is given in Figure 1.
We transform the dataset into a stance detection task by interpreting the question as a naturallanguage representation of the target, and the commentary as the input to be classified.
The dataset is split into a multilingual
training set and into several test sets to evaluate
zeroshot cross-lingual and cross-target transfer. To
provide a baseline, we fine-tune a multilingual
BERT model
We have made the dataset and the code for reproducing the baseline models publicly available.1 Should Switzerland strive for a free trade agreement with the USA? Soll der Bundesrat ein Freihandelsabkommen mit den USA anstreben? La Suisse devrait-elle conclure un accord de libre-échange avec les Etats-Unis? Comment #26597 (German) Label: FAVOR Mit unserem zweitwichtigsten Handelspartner sollten wir ein Freihandelsabkommen haben. [With our second most important trading partner we should have a free trade agreement.] Comment #21421 (French) Label: AGAINST Les accords de libre-échange menacent la qualité des produits suisses. [The free trade agreements jeopardize the quality of the Swiss products.]
Multilingual Stance Detection In the context of the IberEval shared tasks, two related multilingual datasets have been created (Taulé et al., 2017, 2018). Both are a collection of annotated Spanish and Catalan tweets. Crucially, the tweets in both languages focus on the same issue (Catalan independence); given this fact they are the first truly multilingual stance detection datasets known to us.
With regard to the languages covered by
X-stance, only monolingual datasets seem to be
available. For French, a collection of tweets
on French presidential candidates has been
annotated with stance
2016 task on detecting stance in tweets
Sobhani et al. (2017) introduced a multi-target stance dataset which provides two targets per instance. For example, a model designed in this framework is supposed to simultaneously classify a tweet with regard to Clinton and with regard to Trump. While in theory the framework allows for more than two targets, it is still restricted to a finite and clearly defined set of targets. It focuses on modeling the dependencies of multiple targets within the same text sample, while our approach focuses on learning stance detection from many samples with many different targets.
In a target-specific setting, Ghosh et al. (2019)
perform a systematic evaluation of stance
detection approaches. They also evaluate BERT
However, they only experiment with a singlesegment encoding of the input, preventing crosstarget transfer of the model. Augenstein et al. (2016) propose a conditional encoding approach to encode both the target and the tweet as sequences. They use a bidirectional LSTM to condition the encoding of the tweets on the encoding of the target, and then apply a nonlinear projection on
the conditionally encoded tweet. This allows them to train a model that can generalize to previously unseen targets. 3 3.1
The input provided by X-stance is two-fold: (A) a natural language question concerning a political issue; (B) a natural language commentary on a specific stance towards the question.
The label to be predicted is either ‘favor’ or ‘against‘. This corresponds to a standard established by Mohammad et al. (2016a). However, X-stance differs from that dataset in that it lacks a ‘neither’ class; all comments refer to either a ‘favor’ or an ‘against‘ position. The task posed by X-stance is thus a binary classification task.
As an evaluation metric we report the macroaverage of the F1-score for ‘favor’ and the F1score for ‘against’, similar to Mohammad et al. (2016b). We use this metric mainly to strengthen comparability with the previous benchmarks. 3.2
Provenance We downloaded the questions and answers via the Smartvote API2. The downloaded data cover 175 communal, cantonal and national elections between 2011 and 2020.
All candidates in an election who participate in Smartvote are asked the same set of questions, but 2https://smartvote.ch depending on the locale they see translated versions of the questions. They can answer each question with either ‘yes’, ‘rather yes’, ‘rather no’, or ‘no’. They can supplement each answer with a comment of at most 500 characters.
The questions asked on Smartvote have been edited by a team of political scientists. They are intended to cover a broad range of political issues relevant at the time of the election. A detailed documentation of the design of Smartvote and the editing process of the questions is provided by Thurman and Gasser (2009).
Preprocessing We merged the two labels on each pole into a single label: ‘yes’ and ‘rather yes’ were combined into ‘favor’; ‘rather no’, or ‘no’ into ‘against‘. This improves the consistency of the data and the comparability to previous stance detection datasets. We did not further preprocess the text of the comments.
Language Identification As the API does not
provide the language of comments, we employed
a language identifier to automatically annotate
this information. We used the langdetect
library
We applied the identifier in a two-step approach. In the first run we allowed the identifier to output all 55 languages that it supports out of the box, plus Romansh, the fourth official language in Switzerland3. We found that no Romansh comments were detected and that all unexpected outputs were misclassifications of German, French or Italian comments. We further concluded that little or no Swiss German comments are in the dataset; otherwise, some of them would have manifested themselves via misclassifications (e.g. as Dutch).
In the second run, drawing from these conclusions, we restricted the identifier’s set of choices to English, French, German and Italian.
Filtering We pre-filtered the questions and answers to improve the quality of the dataset. To keep the domain of the data surveyable, we set a focus on national-level questions. Therefore, all 3Namely the Rumantsch Grischun variety; the language profile was created using resources from the Zurich Parallel Corpus Collection (Graën et al., 2019) and the Quotidiana corpus (https://github.com/ ProSvizraRumantscha/corpora).
DE FR IT
(New answers to known questions)
Test: Valid:
Test: Valid: Test: questions and corresponding answers pertaining to national elections were included.
In the context of communal and cantonal elections, candidates have answered both local questions and a subset of the national questions. Of those elections, we only considered answers to the questions that also had been asked in a national election. They were only used to augment the training set while the validation and test sets were restricted to answers from national elections.
We discarded the fewer than 20 comments classified as English. Furthermore, we discarded instances that met any of the following conditions: Question is not a closed question or does not address a clearly defined political issue.
No comment was submitted by the candidate or the comment is shorter than 50 characters. Comment starts with “but” or a similar indicator that the comment is not self-contained.
In total, a fifth of the comments were filtered out. Topics The questions have been organized by the Smartvote editors into categories (such as “Economy”). We further consolidated the predefined categories into 12 broad topics (Table 1). Compliance The dataset is shared under a CC BY-NC 4.0 license. Copyright remains with www.smartvote.ch.
Given the sensitive nature of the data, we
increase the anonymity of the data by hashing the
respondents’ IDs. No personal attributes of the
respondents are included in the dataset. We provide
a data statement
We held out the topics “Healthcare” and “Political System” from the training data and created a separate cross-topic test set that contains the questions and answers related to those topics.
Furthermore, in order to test cross-question generalization performance within previously seen topics, we manually selected 16 held-out questions that are distributed over the remaining 10 topics. We selected the held-out questions manually because we wanted to make sure that they are truly unseen and that no paraphrases of the questions are found in the training set.
We designated Italian as a test-only language, since relatively few comments have been written in Italian. From the remaining German and French data we randomly selected a percentage of respondents as validation or as test respondents.
As a result we received one training set, one validation set and four test sets. The sizes of the sets are listed in Table 2. We did not consider test sets that are cross-lingual and cross-target at the same time, as they would have been too small to yield significant results.
Class Distribution Figure 2 visualizes the proportion of ‘favor’ and ‘against‘ stances for each target in the dataset. The ratio differs between questions but is relatively equally distributed across the topics. In particular, the questions in the held-out topics (with a ‘favor’ ratio of 49.4%) have a similar class distribution as the questions in other topics (with a ‘favor’ ratio of 50.0%). Linguistic Properties Not every question is unique; some questions are paraphrases describing the same political issue. For example, in the 2015 election, the candidates were asked: “Should the consumption of cannabis as well as its possession for personal use be legalised?” Four years later they were asked: “Should cannabis use be legalized?” However, we do not see any need to consolidate those duplicates because they contribute to the diversity of the training data.
We further observe that while some questions in the dataset are quite short, some questions are rather convoluted. For example, a typical long question reads:
Some 1% of direct payments to Swiss agriculture currently go to organic farming operations. Should this proportion be increased at the expense of standard farming operations as part of Switzerland’s 2014-2017 agricultural policy? Such longer questions might be more challenging to process semantically.
Languages The X-stance dataset has more
German samples than French samples. The language
ratio of about 3:1 is consistent across all
training and test sets. Given the two languages it
is possible to either train two monolingual
models or to train a single model in a multi-source
setup
We evaluate four baselines to obtain an impression of the difficulty of the task. 4.1
The first pair of baselines uses the most frequent class in the training set for prediction. Specifically, the global majority class baseline predicts the most frequent class across all training targets while the target-wise majority class baseline predicts the class that is most frequent for a given target question. The latter can only be applied to the intra-target test sets. 4.2
As a second baseline, we train a fastText
bag-ofwords linear classifier
The ‘against’ class was slightly upsampled in the training data so that the classes are balanced when summing over all questions and topics.
We use the standard settings provided by the fastText library.4 Optimal hyperparameters from the following range were determined based on the validation accuracy:
Learning rate: 0.1, 0.2, 1
The word vectors were set to a size of 300. We do not initialize them with pre-trained multilingual embeddings since preliminary experiments did not show a beneficial effect. 4.3
As our main baseline model we fine-tune
multilingual BERT (M-BERT) on the task
We fine-tune the full model with a cross-entropy
loss, using the AllenNLP library
Learning rate: 5e-5, 3e-5, 2e-5
4https://github.com/facebookresearch/ fastText
5https://github.com/google-research/ bert/blob/master/multilingual.md
34.8 65.1 71.2 76.6
IT
The grid search was repeated independently for every variant that we test in the following subsections. Furthermore, the standard recommendations for fine-tuning BERT were used: Adam with 1 = 0:9 and 2 = 0:999; an L2 weight decay of 0:01; a learning rate warmup over the first 10% of the steps; and a linear decay of the learning rate. A dropout probability of 0.1 was set on all layers. Results Table 3 shows the results for the crosslingual setting. M-BERT performs consistently better than the previous baselines. Even the zeroshot performance in Italian, while significantly lower than the supervised scores, is much better than the target-wise majority class baseline.
Results for the cross-target setting are given in Table 4. Similar to the cross-lingual setting, model performance drops in the cross-target setting, but M-BERT remains the strongest baseline and easily surpasses the majority class baselines. Furthermore, the cross-question score of M-BERT is slightly lower than the cross-topic score. 4.4
The default setup preserves horizontal language consistency in that the language of the questions always corresponds to the language of the comments. For example, the Italian test instances are combined with the Italian version of the questions, even though during training the model has only ever seen the German and French version of them.
An alternative concept is vertical language consistency, whereby the questions are consistently presented in one language, regardless of the comment. To test whether horizontal or vertical consistency is more helpful, we train and evaluate M-BERT on a dataset variant where all questions are in their English version. We chose English as a lingua franca because it had the largest share of data during the pre-training of M-BERT.
DE FR In order to rule out that only the questions or only the comments are necessary to optimally solve the task, we conduct some additional experiments: Only use a single segment containing the comment, removing the questions from the training and test data (missing questions). Only use the question and remove the comment (missing comments).
In both cases the performance decreases across all evaluation settings (Table 5). The loss in performance is much higher when comments are missing, indicating that the comments contain the most important information about stance. As can be expected, the score achieved without comments is only slightly different from the target-wise majority class baseline.
But there is also a loss in performance when the questions are missing, which underlines the importance of pairing both pieces of text. The effect of missing questions is especially strong in the supervised and cross-lingual settings. To illustrate this, we provide in Table A8 some examples of comments that occur with multiple different targets in the training set. Those examples can explain why the target can be essential for disambiguating a stance detection problem. On the other hand, the effect of omitting the questions is less pronounced in the cross-target settings.
The above single-segment experiments tell us that both the comment and the question provide crucial information. But it is possible that the M-BERT model, even though trained on both segments, mainly looks at a single segment at test time. To rule this out, we probe the model with randomized data at test time:
Test the model on versions of the test sets where the comments remain in place but the questions are shuffled randomly (random questions). We make sure that the random questions come from the same test set and language as the original questions.
Keep the questions in place and randomize the comments (random comments). Again we shuffle the comments only within test set boundaries.
The results in Table 5 show that the performance of the model decreases in both cases, confirming that it learns to take into account both segments. 4.6
Finally we test whether the target really needs to be represented by natural language (e.g. “Do you support X?”). An alternative is to represent the target with a trainable embedding instead.
In order to fit target embeddings smoothly into our architecture, we represent each target type with a different reserved symbol from the M-BERT vocabulary. Segment A is then set to this symbol instead of a natural language question.
The results for this experiment are listed in the bottom row of Table 5. An M-BERT model that learns target embeddings instead of encoding a question performs clearly worse in the supervised and cross-lingual settings. From this we conclude that spelled-out natural language questions provide important linguistic detail that can help in stance detection. 5
Our experiments show that M-BERT achieves a reasonable accuracy on X-stance, outperforming majority class baselines and a fastText classifier. M-BERT — with English questions — with missing questions — with missing comments — with random questions — with random comments — with target embeddings Supervised
To put the supervised score into context we list
scores that variants of BERT have achieved on
other stance detection datasets in Table 6. It seems
that the supervised part of X-stance has a similar
difficulty as the SemEval-2016
On the other hand, in the cross-lingual and cross-target settings, the mean score drops by 6–8 percentage points compared to the supervised setting; while zero-shot transfer is possible to a degree, it can still be improved.
The additional experiments (Table 5) validate the results and show that the sequence-pair classification approach to stance detection is justified.
It is interesting to see what errors the M-BERT model makes. Table A7 presents instances where it predicts the wrong label with a high confidence. These examples indicate that many comments express their stance only on a very implicit level, and thus hint at a potential weakness of the dataset. Because on the voting advice platform the label is explicitly shown to readers in addition to the comments, the comments do not need to express the stance explicitly.
Manual annotation could eliminate very implicit samples in a future version of the dataset. However, the sheer size and breadth of the dataset could not realistically be achieved with manual annotation, and, in our view, largely compensates for the implicitness of the texts. 6
We have presented a new dataset for political stance detection called X-stance. The dataset extends over a broad range of topics and issues regarding national Swiss politics. This diversity of topics opens up an opportunity to further study multi-target learning. Moreover, being partly Swiss Standard German, partly French and Italian, the dataset promotes a multilingual approach to stance detection.
By compiling formal commentary by politicians on political questions, we add a new text genre to the field of stance detection. We also propose a question–answer format that allows us to condition stance detection models on a target naturally.
Our baseline results with multilingual BERT show that the model has some capability to perform zero-shot transfer to unseen languages and to unseen targets (both within a topic and to unseen topics). However, there is some gap in performance that future work could address. We expect that the X-stance dataset could furthermore be a valuable resource for fields such as argument mining, argument search or topic classification.
This work was funded by the Swiss National Science Foundation (project MUTAMUR; no. 176727). We would like to thank Isabelle Augenstein, Anne Göhring and the anonymous reviewers for helpful feedback.
Question Befürworten Sie eine vollständige Liberalisierung der Geschäftsöffnungszeiten? [Are you in favour of a complete liberalisation of business hours for shops?] Soll die Schweiz innerhalb der nächsten vier Jahre EU-Beitrittsverhandlungen aufnehmen? [Should Switzerland embark on negotiations in the next four years to join the EU?] Befürworten Sie einen Ausbau des Landschaftsschutzes? [Are you in favour of extending landscape protection?] La Suisse devrait-elle engager des négociations pour un accord de libre échange avec les EtatsUnis? [Should Switzerland start negotiations with the USA on a free trade agreement?]
Comment Ausser Sonntag. Dies sollte ein Ruhetag bleiben können. [Except Sunday. That should remain a day of rest.] In den nächsten vier Jahren ist dies wohl unrealistisch. [For the next four years this is probably unrealistic.] Wenn es darum geht erneuerbare Energien zu fördern, ist sogar eine Lockerung angebracht. [When it comes to promoting renewable energy, even a relaxation is appropriate.] Il faut cependant en parallèle veiller à ce que la Suisse ne soit pas mise de côté par les Etats-Unis ! [At the same time it must be ensured that Switzerland is not sidelined by the United States!]
Gold Label FAVOR
Prob. [The principle of personal responsibility and corporate self-regulation applies here!] is favorable towards target . . .
but against target . . .
Soll die Schweiz mit den USA Verhandlungen über ein Freihandelsabkommen aufnehmen? [Should Switzerland start negotiations with the USA on a free trade agreement?] Sind Sie für eine vollständige Liberalisierung der Ladenöffnungszeiten? [Are you in favour of the complete liberalization of shop opening times?]
Soll die Schweiz das SchengenAbkommen mit der EU kündigen und wieder verstärkte Personenkontrollen direkt an der Grenze einführen? [Should Switzerland terminate the Schengen Agreement with the EU and reintroduce increased identity checks directly on the border?] Würden Sie die Einführung einer Frauenquote in Verwaltungsräten börsenkotierter Unternehmen befürworten? [Would you support the introduction of a woman’s quota for the Boards of Directors of listed companies?] Table A8: Two comments that imply a positive stance towards one target issue but a negative stance towards another target issue. Such cases can be found in the dataset because respondents have copy-pasted some comments. These examples have been extracted from the training set. Curation rationale In order to study the automatic detection of stances on political issues, questions and candidate responses on the voting advice application smartvote.ch were downloaded. Mainly data pertaining to national-level issues were included to reduce variability.
Language variety The training set consists of questions and answers in Swiss Standard German and Swiss French (74.1% de-CH; 25.9% fr-CH). The test sets also contain questions and answers in Swiss Italian (67.1% de-CH; 24.7% fr-CH; 8.2% it-CH). The questions have also been translated into English.
Candidates for communal, cantonal or national elections in Switzerland who have filled out an online questionnaire.
Socioeconomic status: Unknown – mixed.
The questions were edited and translated by political scientists for a public voting advice website.
The answers were written between 2011 and 2020 by the users of the website.