Hope speech detection in Spanish: The LGBT case

Area of Science:

• Computational Linguistics
• Social Computing
• Natural Language Processing

Background:

• Online content moderation increasingly focuses on detecting abusive and hateful language.
• Existing methods primarily target negative speech, leaving a gap in promoting positive online discourse.
• Hope speech, defined as language that alleviates distress and inspires positivity, offers a counter-narrative to online negativity.

Purpose of the Study:

• To conduct a comprehensive review of existing hope speech detection solutions and resources.
• To introduce and describe the SpanishHopeEDI dataset, a novel resource for Spanish-language hope speech analysis focused on the LGBT community.
• To establish baseline experimental results for future research in automated hope speech detection.

Main Methods:

• Literature review of current hate speech, offensive language, and abusive language detection systems.
• Dataset creation: SpanishHopeEDI, a curated Twitter dataset specifically for the LGBT community.
• Experimental analysis to provide baseline performance metrics for hope speech detection.

Main Results:

• A thorough analysis of existing approaches to online content moderation and hope speech.
• The creation and release of SpanishHopeEDI, a valuable resource for studying positive online communication within a specific community.
• Initial experimental findings that set a benchmark for future research in this domain.

Conclusions:

• Automated detection and promotion of hope speech can significantly combat discrimination and foster healthier online environments.
• The SpanishHopeEDI dataset represents a crucial step forward in enabling research on hope speech, particularly concerning marginalized communities.
• Further research is warranted to build upon these findings and develop more robust hope speech detection systems.