> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/ItzCrazyKns/Perplexica/llms.txt
> Use this file to discover all available pages before exploring further.

# Search sources

> Select specific sources to search across the web, academic papers, or community discussions

Perplexica allows you to choose which sources to search, helping you find the most relevant information for your query. Different sources are optimized for different types of questions.

## Available sources

<CardGroup cols={3}>
  <Card title="Web" icon="globe">
    General web search across all websites
  </Card>

  <Card title="Academic" icon="graduation-cap">
    Scholarly articles and research papers
  </Card>

  <Card title="Discussions" icon="comments">
    Forums, reviews, and community content
  </Card>
</CardGroup>

## Web search

Web search queries the general internet for information across all types of websites. This is the default source and works well for most queries.

**Best for:**

* General information and news
* How-to guides and tutorials
* Product information
* Current events
* Company and organization information

**What it searches:**

Web search uses SearxNG to query multiple search engines simultaneously while maintaining your privacy. Results include articles, blog posts, documentation, news sites, and general web content.

<Info>
  Perplexica's web search is powered by SearxNG, which aggregates results from multiple search engines without tracking your queries.
</Info>

## Academic search

Academic search focuses on scholarly articles, research papers, and academic publications.

**Best for:**

* Scientific research
* Academic citations
* Peer-reviewed studies
* Technical and scholarly information
* Research methodology

**What it searches:**

Academic mode prioritizes results from academic databases, university websites, research journals, and scholarly publications. Citations come from credible academic sources.

<Note>
  When academic search is selected, the classifier enables `academicSearch` mode, which adjusts the research strategy to prioritize scholarly sources.
</Note>

**Example queries:**

* "Latest research on climate change mitigation"
* "Machine learning applications in healthcare"
* "Economic impact of renewable energy adoption"

## Discussion search

Discussion search finds information from forums, community discussions, user reviews, and Q\&A sites.

**Best for:**

* Product reviews and user experiences
* Troubleshooting and problem-solving
* Community opinions and feedback
* Real-world usage examples
* Comparisons based on user experiences

**What it searches:**

Discussion mode targets forums like Reddit, Stack Overflow, product review sites, Q\&A platforms, and community discussion boards. You get insights from real users sharing their experiences.

<Tip>
  Discussion search is particularly valuable when you want to hear from actual users rather than official documentation or marketing materials.
</Tip>

**Example queries:**

* "Best laptop for programming Reddit"
* "Common issues with \[product name]"
* "User experiences with \[service]"

## Combining sources

You can select multiple sources simultaneously to get a broader perspective:

* **Web + Academic**: General information supplemented with research findings
* **Web + Discussions**: Official information plus user experiences
* **All three**: Comprehensive coverage from all angles

## How source selection works

When you select sources in the UI, they're passed to the search agent:

```typescript theme={null}
sources: SearchSources[] // ['web', 'discussions', 'academic']
```

The classifier then determines which research actions to enable:

```typescript theme={null}
classification: {
  skipSearch: boolean;
  personalSearch: boolean;
  academicSearch: boolean;
  discussionSearch: boolean;
  // ...
}
```

## Search engine privacy

Perplexica uses SearxNG as the default search backend, which provides several privacy benefits:

* No tracking of your search queries
* No personalization based on search history
* Aggregates results from multiple engines
* Removes tracking parameters from result URLs

<Info>
  You can configure your own SearxNG instance by setting the `SEARXNG_API_URL` environment variable. Make sure JSON format is enabled in your SearxNG settings.
</Info>

## Domain-specific search

In addition to source types, you can limit searches to specific domains:

* Perfect for searching within documentation sites
* Useful for research within specific publications
* Great for finding content on trusted sources

Simply include `site:example.com` in your query to search within a specific domain.

## Upcoming sources

The following search sources are planned for future releases:

* **Tavily**: Enhanced search API optimized for LLMs
* **Exa**: Neural search for better semantic understanding
* Additional specialized search engines