In the current scientific research environment, where quantity is paramount and blindly following trends is the norm, papers that cannot be reproduced, are futile, and boastful abound. Many seemingly brilliant ideas, after peeling back the surface, are just clever tricks on specific datasets; those glorious metrics and breakthroughs are often just flashes in the pan under experimental settings. Future generations will run wildly along these "milestones," only to find that the foundation is made of papier-mâché midway: they will fall into self-doubt, and in the end, they can only laugh bitterly, treating this experience as black humor on the road of scientific research.
DisdainAbstract was born in such an absurd atmosphere. It does not intend to solve problems, nor does it pretend to be constructive; it only wants to use a playful tone to expose the "old wine in new bottles" routine, the "experimental miracles" piled up by data manipulation, and the seemingly profound but empty module designs in theses. It imitates the circle of those poisonous predecessors, translating "novel" into "skinning," "extensive experiments" into "metric acrobatics," "significant improvement" into "the art of overfitting." It doesn't expect to change anything, but only intends to hand you a sentence when you are changing code late at night and wondering if you are also producing academic rubbish:
- This tool is for academic analysis and entertainment purposes only. Do not use it to maliciously attack others' work.
- API calls may incur costs; please check the pricing of the respective company.
- Ensure compliance with academic ethics and copyright regulations.
- Node.js (v14 or higher)
- npm or yarn
-
Clone the repository:
git clone https://github.com/your-username/disdain-abstract.git cd disdain-abstract -
Install dependencies:
npm install
-
Configure API Key:
- Open
src/server.js - Replace
KIMI_API_KEYwith your Moonshot Kimi API Key, or use another LLM API.
- Open
-
Download fonts (optional, for best display):
.\scripts\download-jetbrains-mono.ps1
npm startThe server will start at http://localhost:3000. Open your browser and visit that address to use the application.
- Enter or paste an academic abstract in the text box, or input an arXiv URL.
- Select an analysis style: Sarcastic, Angry, Weary.
- You can add custom bias to influence the analysis results.
- Tired of serious academic language? Go Crazy to turn it into nonsense.