What is Perplexica
Perplexica is an open-source AI-driven search engine and an open-source alternative to Perplexity AI. It uses machine learning algorithms and natural language processing technology to understand user queries and provide precise answers. Perplexica supports multiple search modes, including web search, writing assistant, academic search, YouTube search, Wolfram Alpha search, and Reddit search. Perplexica supports local large language models, uses APIs to integrate search functionality into other applications, and integrates SearxNG search technology to ensure access to the latest information.
Main Features of Perplexica
- Local LLM Support: Perplexica supports the use of local large language models (LLMs), such as Llama3 and Mixtral, to improve search accuracy.
- Normal Mode: Processes user queries and performs web searches.
- Focus Mode: Includes multiple specific types of search modes, such as:
- Web Search Mode: Searches the entire web to find the best results.
- Writing Assistant Mode: Helps with writing tasks that do not require web search.
- Academic Search Mode: Searches academic articles and papers, suitable for academic research.
- YouTube Search Mode: Finds YouTube videos based on the query.
- Wolfram Alpha Search Mode: Uses Wolfram Alpha for queries that require computation or data analysis.
- Reddit Search Mode: Searches Reddit to find discussions and opinions related to the query.
- API Integration: Supports developers in integrating Perplexica into their own applications to use the search functionality.
- Real-time Information Updates: Uses the SearxNG meta search engine to ensure the latest information is provided.
Technical Principles of Perplexica
- User Interaction: The user sends messages to the backend server based on WebSocket, triggering the processing chain.
- Query Processing: The message is passed to a chain, which predicts whether a web search is needed based on the chat history and the question. If needed, a query will be generated.
- Web Search: The query performs a web search based on SearxNG to obtain information.
- Similarity Search: The searched information is converted into embeddings and a similarity search is performed to find the most relevant resources.
- Response Generation: The resources are passed to the response generator, which generates a response by combining the chat history, query, and resources.
- User Interface Display: The final response is displayed to the user based on the user interface, completing the entire search and response process.
Perplexica Project Address
- GitHub Repository：https://github.com/ItzCrazyKns/Perplexica
Application Scenarios of Perplexica
- Personal Search Engine: Users can use it as a personal search engine to protect privacy and customize the search experience.
- Academic Research: Researchers can find academic papers and articles to support research work.
- Enterprise Internal Search: Enterprises can build a search engine for internal knowledge bases to help employees quickly find the information they need.
- Developer Tools: Developers integrate Perplexica's API into their own applications to enhance the search capabilities of the apps.
- Educational purposes: Educational institutions help students and teachers quickly access educational resources and academic materials.