This paper describes RMIT-IR team's participation in the EXIST Lab at CLEF 2024. The proposed approaches aim to address sexism characterization on microblog posts (Tasks 1, 2, and 3) and sexism identification on memes (Task 4). For Tasks 1-3, we studied the efectiveness of zero-shot In-Context Learning (ICL) [1] with of-the-shelf pre-trained Large Language Models (LLMs) to mimic the scenario of minimal intervention of a practitioner aiming to build sexism characterization systems. Our approaches for meme classification (Task 4) utilize CLIP (Contrastive Language-Image Pre-training) [ 2] to experiment with multi-modal embeddings and zero-shot sexism identification models. We report the performance of our approaches under the learning with disagreements regime (Soft evaluation) and also for label predictions (Hard evaluation). The code of our submission is available at https://github.com/rmit-ir/exist2024/.
CEUR ceur-ws.org
Social media has had a considerable impact on human societies. Applications such as Facebook,
YouTube, Instagram, and TikTok have helped move the zeitgeist while creating large
communities in the millions. However, many social media platforms have issues with people creating and
posting harmful information. The challenge of detecting and managing such harmful content
remains a concern for social media companies, contributing to consequences ranging from
misinformation to adverse efects on mental health [
Sexism is the belief that the members of one sex or gender are less than the members of the
other sex, especially that women are less able than men [
Sexism can limit the opportunities and roles people of diferent sexes and genders are expected
to take. It can be conveyed through any form of expression, like images, cartoons, memes,
objects, gestures, and symbols, and can be spread ofline or online. This oppression can take
diferent forms, such as economic exploitation and social domination [
Sexist attitudes and behaviours can perpetuate stereotypes of social and gender roles based
on one’s biological sex. Usually, people are socialized with sexist concepts that teach traditional
gender roles for males and females [
There has been a recent increase in research on identifying diferent forms of hate speech,
corresponding with advancements in generative pre-trained transformers and, in general, large
language models (LLMs) [
In addition to the text classification in Tasks 1–3, we address the problem of identifying
sexism in multi-modal formats for Task 4. Memes — ideas, images, or videos that are spread
very quickly on the internet [
The paper is organized as follows. Details about the tasks participated in are described in Section 2. Section 3 provides details about the proposed approaches. In Section 4, we provide and discuss the results. Finally, we conclude in Section 5.
The sEXism Identification in Social neTworks (EXIST) [
• Task 1: Addresses sexism identification in tweets as a binary classification, requiring the system to classify whether a tweet is sexist (YES) or not (NO). • Task 2: Focuses on determining the source intention in tweets as a multi-class classification, requiring the system to classify the tweet’s intention as Direct, Reported, or Judgemental. • Task 3: Involves sexism categorization in tweets as a multi-label classification, requiring the system to classify tweets into categories such as Ideological Inequality, Stereotyping Dominance, Objectification , Sexual Violence, and Misogyny-Non-Sexual Violence.
While the above tasks address sexism identification in text, Task 4 deals with multi-modal input. Task 4 aims to address sexism identification as a binary classification, requiring the systems to classify whether a given meme is sexist or not.
Task Guidelines
Prompt
Prompt Instantiation
Pre-trained Large Language
Model (e.g.,
GPT-4) /
Activity with / without manual intervention
Artifact LLM Output
Valid
Run Output Validation
• Soft-Soft Evaluation: For systems that produce probabilities for each category, soft-soft
evaluation is provided to compare the probabilities assigned by the systems with those
assigned by the set of human annotators. The oficial evaluation metric is ICM-soft [
Additionally, Cross Entropy is also reported. • Hard-Hard Evaluation: Hard labels are derived from the diferent annotators’ labels through a probabilistic threshold computed for each task. Hard-hard evaluation is provided to evaluate systems that return Hard labels as output by comparing against a ground truth that combines multiple annotations into one. The original ICM [25] and 1 score are used as evaluation metrics.
Our goal was to examine the procedure of developing a functional solution with readily available LLMs while minimizing the manual efort required from the practitioner. As shown in Figure 1, the basic architecture involves giving the researcher a set of labeling or classification tasks and asking the LLM to generate an accurate output. To ensure that the responses followed the predefined criteria for each task, the outputs were systematically stored in a JSON format and manually inspected for errors in the ”value” field. Responses such as ”YES”, ”YES”, or variations with additional text or punctuation like ”Yes, the ... is sexist” required manual corrections to conform to the expected format. Incidences of token limit rates that resulted in ”HTTP” errors were addressed by re-running the task for the afected tweet using its unique ID. These occurrences were uncommon, making manual correction a more eficient solution than an automated task given the time constraint.
The prompts used for runs submitted to Tasks 1, 2, and 3 were designed with multiple parts: • Definition of the underlying concept being addressed in the task (e.g., sexism): Sexism, prejudice or discrimination based on sex or gender, especially against women and girls. Although its origin is unclear, the term sexism emerged from the “second-wave” feminism of the 1960s through ‘80s and was most likely modeled on the civil rights movement’s term racism (prejudice or discrimination based on race). Sexism can be a belief that one sex is superior to or more valuable than another sex. It imposes limits on what men and boys can and should do and what women and girls can and should do. The concept of sexism was originally formulated to raise consciousness about the oppression of girls and women, although by the early 21st century it had sometimes been expanded to include the oppression of any sex, including men and boys, intersex people, and transgender people. • Instruction to Address Task and to Obtain Consistent Outputs: You are a robot who detects sexism from text given in the prompt. • Perspectivism: – Level of Education: For each response, consider the perspective of individuals representing the following study levels: [study_levels_annotators] – Level of Education and Gender: For each response, consider the perspective of individuals representing the following study levels: [study_levels_annotators] and gender: [gender_annotators]. • Output Format: – Task 1 (Soft): Give me 6 answers with NO or YES. Format: [NO], [YES] – Task 1 (Hard): Give me 1 answer with [NO] or [YES] – Task 2 (Soft) : Give me 6 answers with NO, DIRECT, REPORTED or JUDGEMENTAL using commas for each answer. Example: [NO], [DIRECT], [REPORTED], [JUDGEMENTAL], [JUDGEMENTAL], [NO] – Task 2 (Hard): Give me 1 answer with NO, DIRECT, REPORTED or JUDGEMENTAL using commas for each answer. Example: [NO], [DIRECT], [REPORTED], [JUDGEMENTAL], [JUDGEMENTAL], [NO] – Task 3 (Soft): Give me 6 answers with NO, IDEOLOGICAL-INEQUALITY, STEREOTYPING-DOMINANCE, OBJECTIFICATION, SEXUAL-VIOLENCE, or MISOGYNY-NON-SEXUAL-VIOLENCE using commas for each answer. Example: [NO], [IDEOLOGICAL-INEQUALITY], [STEREOTYPING-DOMINANCE], [OBJECTIFICATION], [SEXUAL-VIOLENCE], [MISOGYNY-NON-SEXUAL-VIOLENCE] – Task 3 (Hard): Give me 1 answers with NO, IDEOLOGICAL-INEQUALITY, STEREOTYPING-DOMINANCE, OBJECTIFICATION, SEXUAL-VIOLENCE, or MISOGYNY-NON-SEXUAL-VIOLENCE using commas for each answer. Example: [NO], [IDEOLOGICAL-INEQUALITY], [STEREOTYPING-DOMINANCE], [OBJECTIFICATION], [SEXUAL-VIOLENCE], [MISOGYNY-NON-SEXUAL-VIOLENCE] • Instance to Classify: #### [tweet] #### An example of a prompt submitted and output obtained from gpt-4-turbo to classify instance with id_EXIST 600090 using RMIT-IR_3 for Task 2 (Soft): • Input: Sexism, prejudice or discrimination based on sex or gender, especially against women and girls. Although its origin is unclear, the term sexism emerged from the “second-wave” feminism of the 1960s through ‘80s and was most likely modeled on the civil rights movement’s term racism (prejudice or discrimination based on race). Sexism can be a belief that one sex is superior to or more valuable than another sex. It imposes limits on what men and boys can and should do and what women and girls can and should do. The concept of sexism was originally formulated to raise consciousness about the oppression of girls and women, although by the early 21st century it had sometimes been expanded to include the oppression of any sex, including men and boys, intersex people, and transgender people. You are a robot who detects sexism from text given in the prompt. For each response, consider the perspective of individuals representing the following study levels: [“High school degree or equivalent”, “Bachelor’s degree”, “Bachelor’s degree”, “Bachelor’s degree”, “Bachelor’s degree”, “High school degree or equivalent”]. Give me 6 answers with NO, DIRECT, REPORTED or JUDGEMENTAL using commas for each answer. Example: [NO], [DIRECT], [REPORTED], [JUDGEMENTAL], [JUDGEMENTAL], [NO]. #### Girls, don’t let anyone ever tell you, you’re not as good as a man #gender #girlpower #equity #### • Output: [NO],[NO],[NO],[NO],[NO],[NO] To find the distribution of responses, we initially tried to use GPT to figure out the likelihood percentage. Unfortunately, GPT only gave absolute values (either 100 or 0) or a consistent split of 70/30 most of the time. We directed the model to generate six responses for each tweet, which matched the number of annotators per tweet. For example, a set of responses like ”YES”, ”NO”, ”YES”, ”NO”, ”YES”, and ”NO” would result in a calculated distribution of 50%. Additionally, in our final submissions, experimental runs two and three included prompts that provided additional context, such as the annotators’ gender or educational backgrounds. This was done to see if providing relevant background information would improve the LLM’s ability to predict annotator responses. The formats for each task can be seen above along with an example prompt used in RMIT-IR_3 for Task 2 (Soft).
We used OpenAI’s API to submit prompts to the pre-trained model gpt-4-turbo-2024-0409 [26]. For each tweet in the test set, we instantiated the prompt from above by appending the textual content of the instance. We used the syntax #### [tweet] #### to provide explicit delimiters to the model. For Soft tasks, we asked for six instances and then created a distribution based on the frequency of the predicted labels.
We experimented with multiple versions of prompt templates using the development set supplied by the EXIST organizers (we did not use the training set). We found that the following elements were especially efective in directing the model to concentrate on the specific task and to ensure the responses were properly formatted (i.e., single-word answers and capitalized): • Employing a role-playing technique of framing the task with the prompt “You are a robot who detects sexism from text given in the prompt.” • Giving explicit formatting instructions such as “Give me 6 answers with NO or YES. Format: [NO], [YES]”.
Inspired by the successful applications of CLIP [
We propose two supervised contrastive learning models based on CLIP: Text-Image multimodal model via CLIP-guided learning (TI-CLIP) and Text-Image Multi-View multi-modal model via CLIP-guided learning (TIMV-CLIP). The architecture of TIMV-CLIP is shown in Figure 2. We also propose Prompt-CLIP to address zero-shot sexism classification and CLIP-based models for supervised sexism classification.
• TI-CLIP: The overall architecture of TI-CLIP consists of two feature encoding models
used to encode texts and images. These embeddings are then combined into a multi-modal
embedding before passing into a feedforward network for sexism classification.
• TIMV-CLIP: We adopted a novel multi-view CLIP framework (MV-CLIP) [
Image Encoder
(CLIP) Image Encoder (CLIP)
+ Text Encoder
(CLIP) Text Encoder (CLIP)
Transformer
Encoder
Key-less Attention
Linear
Softmax Linear
Softmax Linear
Softmax
modality using a transformer encoder, which aims to capture the interaction across
diferent modalities. Unlike MV-CLIP in a previous study [
The proposed multi-modal sexism identification models mainly focus on Soft label predictions. For the Hard submissions, hard labels are directly assigned by applying the max function, i.e., based on the highest probability score.
Table 2 presents the submitted runs for Task 4, which can be summarized as follows: • RMIT-IR_1: For the first submission, the trained TI-CLIP model was used to predict whether given memes are sexist or not sexist. • RMIT-IR_2: We used the trained TIMV-CLIP to generate the second submission. • RMIT-IR_3: Prompt-CLIP was used to predict Soft and Hard labels for the third submission.
4.1. Tasks 1–3 The performance of our proposed approaches for Tasks 1, 2, and 3 are presented in Tables 3, 4, and 5, respectively. As shown in Table 3, simplifying the architecture required for creating a LLM that can identify sexism shows promise, as evidenced by the classification of English and Spanish tweets. Although the current model achieved a 49% ICM-Soft score, which is 17% lower than the best-performing run, this result indicates the potential to use prompting for classification tasks.
The cost of this process, which included testing with a development dataset and submitting with a gold dataset, was close to $150 AUD. In particular, it did not require knowledge of cloud computing, expensive hardware, or much energy. The average time taken to produce an output was about 90 minutes.
Looking at Table 3, Task 1, the use of the Task-specific Prompt yielded an ICM-Soft Norm score of 49%, securing the 23rd position overall. It is interesting to note that the inclusion of additional clues such as the annotator’s education level and gender did not bolster performance; instead, it diminished the score. In particular, when looking at the Spanish test instances, the model scored higher across all Spanish test instances compared to all English test instances, despite being trained on an English-based GPT. This underscores the robust cross-lingual applicability of the model, showcasing its proficient handling of Spanish data despite its primary training on English.
In Task 2, our analysis revealed challenges in multi-class classification as shown in Table 4. The approach yielded a 13% ICM-Soft Norm score, indicating considerable dificulty in discerning the intention of the tweets. The introduction of additional clues, such as the annotator’s education level, generally led to a decline in performance. However, adding gender information resulted in a slight improvement, elevating the score from almost 0% to 3%. The results indicated that the approach performed more efectively in English without additional clues; however, its performance diminished once clues were introduced. Conversely, our analysis demonstrated that the GPT model exhibited greater eficacy with clues in Spanish, suggesting potential advantages in providing contextual information in non-English scenarios.
Task 3 was also challenging with multi-label classification. The initial ICM-Soft Norm score, as shown in Table 5, stood at 11%. Following a similar trend to Task 2, the integration of clues such as the level of education of the annotator, resulted in a reduction in performance to 8%. Subsequently, when both education and gender clues were included, performance decreased further to 4%. The English scores are quite similar to Task 2. Notably from the initial performance on the Spanish dataset exceeded that of Task 2, but declined with the addition of educational clues and further declined with the incorporation of both education and gender clues. This observation underscores a consistent pattern of diminishing returns with the incorporation of more specific annotator information.
Our proposed approaches for the second and third runs involve implementing few-shot and in-context learning. The experiments for these runs were conducted using gpt-4-turbo, and future tests should include gpt-4o along with other pre-trained LLMs to determine their eficacy in this context. Our experiment results show that involving few-shot and in-context learning does not improve model performance on sexism identification in tweets (as shown in Tables 3–5). Although prompting requires less coding and understanding of LLMs, producing the exact desired response 100% of the time was challenging. The prompts had to be carefully designed to ensure that the GPT provided a single and consistent answer, especially when dealing with distribution. Although pre-processing text for an LLM is more complex than pre-processing answers from GPTs, ensuring the response is in the correct format is simpler. Results of the proposed approaches for Task 4 (Soft and Hard).
Rank (Soft) ICM-Soft ICM-Soft Norm Cross Entropy ICM-Hard ICM-Hard Norm F1 YES Another unexpected aspect of this architecture was the ability to assist the GPT with hints. We tested how adding biases by including the annotator’s education level and gender afected its ability to classify or label tweets. This gave insights into how such biases can influence model performance and classification accuracy. 4.2. Task 4 proaches, TIMV-CLIP performs best in all cases (English+Spanish, English, or Spanish test instances) considering both Soft and Hard evaluation scenarios. This indicates the importance of efectively utilizing deep interactions between texts and images of memes with CLIP. Furthermore, TIMV-CLIP achieved the best-performance model (RMIT-IR_2) on English test instances with an ICM-Soft Norm score of 0.4998, ranked first in the leaderboard considering the Soft evaluation (English test instances). This observation confirms the advantages of CLIP for text-image pair classification tasks. However, the performance of TIMV-CLIP has dropped in Spanish test instances, which leads to lower performance in all test instances (Spanish test instances). We believe using a translation component for Spanish text in memes could lead to better overall performance.
This paper proposed unsupervised in-context learning with of-the-shelf pre-trained LLMs to address sexism characterization on microblog posts (Tasks 1, 2, and 3). Dealing with multi-modal inputs, we proposed multi-modal contrastive learning, including Prompt-CLIP, TI-CLIP, and TIMV-CLIP for sexism identification in memes (Task 4).
The results of our experiment demonstrated the efectiveness of TIMV-CLIP under the Learning with Disagreements regime, indicating the need to consider capturing sexism cues from diferent perspectives, including image, text, and image-text interactions.
Future work includes further experimentation with unsupervised In-Context Learning in other tasks or meta-tasks such as MonsterCLEF [31], and the inclusion of machine translation for multi-modal contrastive learning.
This research has been carried out in the unceded lands of the Woi Wurrung and Boon Wurrung peoples of the eastern Kulin Nation. We pay our respects to their Ancestors and Elders, past, present, and emerging. This research is partially supported by the Australian Research Council (ARC, project nr. DE200100064 and CE200100005). 2024 – Conference and Labs of the Evaluation Forum, 2024. [24] L. Plaza, J. Carrillo-de-Albornoz, R. Morante, E. Amigó, J. Gonzalo, D. Spina, P. Rosso, Overview of EXIST 2023 – Learning with Disagreement for Sexism Identification and Characterization (Extended Overview), in: M. Aliannejadi, G. Faggioli, N. Ferro, M. Vlachos (Eds.), Working Notes of CLEF 2023 – Conference and Labs of the Evaluation Forum, 2023, pp. 813–854. URL: https://ceur-ws.org/Vol-3497/paper-070.pdf. [25] E. Amigó, A. Delgado, Evaluating Extreme Hierarchical Multi-label Classification, in: S. Muresan, P. Nakov, A. Villavicencio (Eds.), Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Linguistics, Dublin, Ireland, 2022, pp. 5809–5819. doi:10.18653/v1/2022. acl-long.399. [26] OpenAI, https://help.openai.com/en/articles/8555510-gpt-4-turbo-in-the-openai-api, 2024.
Accessed: 2024-07-04. [27] J. Fu, S. Xu, H. Liu, Y. Liu, N. Xie, C.-C. Wang, J. Liu, Y. Sun, B. Wang, CMA-CLIP: Cross-Modality Attention Clip for Text-Image Classification, in: 2022 IEEE International Conference on Image Processing (ICIP), 2022, pp. 2846–2850. doi:10.1109/ICIP46576. 2022.9897323. [28] M. V. Conde, K. Turgutlu, CLIP-Art: Contrastive Pre-training for Fine-Grained Art Classification, in: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2021, pp. 3951–3955. doi:10.1109/CVPRW53098.2021.00444. [29] J. Devlin, M.-W. Chang, K. Lee, K. Toutanova, BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding, in: J. Burstein, C. Doran, T. Solorio (Eds.), Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), Association for Computational Linguistics, Minneapolis, Minnesota, 2019, pp. 4171–4186. doi:10.18653/v1/N19-1423. [30] T. Wolf, L. Debut, V. Sanh, J. Chaumond, C. Delangue, A. Moi, P. Cistac, T. Rault, R. Louf, M. Funtowicz, J. Davison, S. Shleifer, P. von Platen, C. Ma, Y. Jernite, J. Plu, C. Xu, T. Le Scao, S. Gugger, M. Drame, Q. Lhoest, A. Rush, Transformers: State-of-the-Art Natural Language Processing, in: Q. Liu, D. Schlangen (Eds.), Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Association for Computational Linguistics, Online, 2020, pp. 38–45. doi:10.18653/v1/2020.emnlp-demos.6. [31] N. Ferro, J. Gonzalo, J. Karlgren, H. Müller, MonsterCLEF: One Lab to Rule Them All, 2024.
URL: https://monsterclef.dei.unipd.it/, Accessed: 2024-07-04.