AIvidence Documentation
AI-powered fact checking tool for fighting misinformation
fact_check_engine
FactCheckEngine for analyzing content and checking its factual accuracy.
FactCheckEngine
Class
Main engine for analyzing content and checking its factual accuracy.
Orchestrates the full pipeline from content retrieval to final report generation.
__init__
Method
def __init__(self, model_name: str = 'gpt-4o-mini', api_key: Optional[str] = None, base_url: Optional[str] = None, verbose: bool = False, brave_api_key: str = 'your_brave_search_key_here'):
Initialize the fact check engine.
Parameters
| Parameter | Description |
|---|---|
model_name |
Name of the LLM to use |
api_key |
API key for the language model |
base_url |
Base URL for the language model (for Ollama models) |
verbose |
Whether to enable verbose logging |
brave_api_key |
API key for Brave Search |
analyze_content
Method
def analyze_content(self, source: str, source_type: Optional[str] = None, max_claims: int = 5) -> aividence.models.content_analysis_report.ContentAnalysisReport:
Analyze content from a URL or HTML file for misinformation.
ContentAnalysisReport with analysis details
Parameters
| Parameter | Description |
|---|---|
source |
Source of content (URL or file path) |
source_type |
Type of source ('url' or 'file'). If None, type is auto-detected. |
max_claims |
Maximum number of claims to verify |
Returns
ContentAnalysisReport with analysis details
analyze_website
Method
def analyze_website(self, url: str, max_claims: int = 5) -> aividence.models.content_analysis_report.ContentAnalysisReport:
Legacy method to maintain backward compatibility.
ContentAnalysisReport with analysis details
Parameters
| Parameter | Description |
|---|---|
url |
Website URL to analyze |
max_claims |
Maximum number of claims to verify |
Returns
ContentAnalysisReport with analysis details
domain_analyzer
DomainAnalyzer agent for analyzing website domains and content topics.
DomainAnalyzer
Class
Agent for analyzing website domains and determining their topics,
required expertise, and verification strategies.
__init__
Method
def __init__(self, model_client: aividence.clients.model_client.ModelClient, verbose: bool = False):
Initialize the domain analyzer.
Parameters
| Parameter | Description |
|---|---|
model_client |
ModelClient instance for querying language models |
verbose |
Whether to enable verbose logging |
analyze_domain
Method
def analyze_domain(self, url: str, content: str) -> Dict[str, Any]:
Analyze the website domain and determine its topic, expertise required, and what to verify.
Dictionary with domain analysis results
Parameters
| Parameter | Description |
|---|---|
url |
Website URL |
content |
Website content |
Returns
Dictionary with domain analysis results
claim_extractor
ClaimExtractor agent for extracting verifiable claims from content.
ClaimExtractor
Class
Agent for extracting verifiable factual claims from website content.
__init__
Method
def __init__(self, model_client: aividence.clients.model_client.ModelClient, verbose: bool = False):
Initialize the claim extractor.
Parameters
| Parameter | Description |
|---|---|
model_client |
ModelClient instance for querying language models |
verbose |
Whether to enable verbose logging |
extract_claims
Method
def extract_claims(self, content: str, domain_analysis: Dict[str, Any], max_claims: int = 10) -> List[aividence.models.fact_claim.FactClaim]:
Extract verifiable claims from the website content.
List of FactClaim objects
Parameters
| Parameter | Description |
|---|---|
content |
Website content |
domain_analysis |
Results from domain analysis |
max_claims |
Maximum number of claims to extract |
Returns
List of FactClaim objects
claim_verifier
ClaimVerifier agent for verifying factual claims using web search.
ClaimVerifier
Class
Agent for verifying factual claims by searching for evidence and analyzing results.
__init__
Method
def __init__(self, model_client: aividence.clients.model_client.ModelClient, search_client: aividence.clients.web_search_client.WebSearchClient, verbose: bool = False):
Initialize the claim verifier.
Parameters
| Parameter | Description |
|---|---|
model_client |
ModelClient instance for querying language models |
search_client |
WebSearchClient instance for searching the web |
verbose |
Whether to enable verbose logging |
verify_claim
Method
def verify_claim(self, claim: aividence.models.fact_claim.FactClaim, domain_analysis: Dict[str, Any]) -> aividence.models.fact_check_result.FactCheckResult:
Verify a single claim by searching for evidence and analyzing results.
FactCheckResult with the verification details
Parameters
| Parameter | Description |
|---|---|
claim |
The claim to verify |
domain_analysis |
Domain analysis results for context |
Returns
FactCheckResult with the verification details
content_scraper
ContentScraper client for loading content from websites or HTML files.
ContentScraper
Class
A class to load website content from URLs or HTML files using multiple fallback methods.
__init__
Method
def __init__(self, user_agent=None, timeout=30, verbose=False):
Initialize the scraper with configurable parameters.
load_content
Method
def load_content(self, source, source_type=None):
Load content from either a website URL or an HTML file.
str: The website content as text
Raises:
ValueError: If the content could not be loaded after all attempts
Parameters
| Parameter | Description |
|---|---|
source |
The URL of the website or path to HTML file |
source_type |
Optional type specifier ('url' or 'file'). If None, type is auto-detected. |
Returns
str: The website content as text
load_website_content
Method
def load_website_content(self, url):
Legacy method to maintain backward compatibility.
str: The website content as text
Parameters
| Parameter | Description |
|---|---|
url |
The URL of the website to load |
Returns
str: The website content as text
model_client
ModelClient for handling interactions with language models.
ModelClient
Class
A wrapper class to handle interactions with different language models.
Supports OpenAI, Anthropic, and Ollama models.
__init__
Method
def __init__(self, model_name: str, api_key: Optional[str] = None, base_url: Optional[str] = None, verbose: bool = False):
Initialize the model client.
Parameters
| Parameter | Description |
|---|---|
model_name |
Name of the model to use |
api_key |
API key for the model service |
base_url |
Base URL for the model service (used with Ollama models) |
verbose |
Whether to enable verbose logging |
run
Method
def run(self, prompt: str, system_prompt: Optional[str] = None) -> str:
Run a chat completion.
The model's response as a string
Parameters
| Parameter | Description |
|---|---|
prompt |
The user prompt to send to the model |
system_prompt |
Optional system prompt to set context |
Returns
The model's response as a string
web_search_client
WebSearchClient for searching the web using Brave Search API.
WebSearchClient
Class
Tool for searching the web using Brave Search API with direct HTTP requests
__init__
Method
def __init__(self, api_key: str, max_results: int = 5, verbose: bool = False, timeout: int = 10, max_retries: int = 3, retry_delay: int = 10):
Initialize the web search client.
Parameters
| Parameter | Description |
|---|---|
api_key |
Brave Search API key |
max_results |
Maximum number of results to return |
verbose |
Whether to enable verbose logging |
timeout |
Request timeout in seconds |
max_retries |
Maximum number of retry attempts |
retry_delay |
Delay between retries in seconds |
search
Method
def search(self, query: str) -> List[aividence.models.search_result.SearchResult]:
Perform a search using Brave Search API with timeout handling and retries.
List of SearchResult objects
Parameters
| Parameter | Description |
|---|---|
query |
Search query string |
Returns
List of SearchResult objects
fact_claim
FactClaim model for representing claims to be verified.
FactClaim
Class
A factual claim extracted from content that needs verification.
Attributes:
id: Unique identifier for the claim
claim: The actual claim text
topic: Topic category of the claim
keywords: List of keywords relevant to the claim
importance: Importance score of the claim (1-10)
Inherits from: BaseModel
__init__
Method
def __init__(self, /, **data: 'Any') -> 'None':
Create a new model by parsing and validating input data from keyword arguments.
Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
validated to form a valid model.
`self` is explicitly positional-only to allow `self` as a field name.
search_result
SearchResult model for representing search results.
SearchResult
Class
Represents a search result from a web search.
Attributes:
title: Title of the search result
body: Body text or description of the search result
url: URL of the search result
source: Source domain of the search result
__init__
Method
def __init__(self, title: str, body: str, url: str, source: str):
Initialize self. See help(type(self)) for accurate signature.
fact_check_result
FactCheckResult model for representing verification results.
FactCheckResult
Class
Represents the result of verifying a factual claim.
Attributes:
claim_id: ID of the claim being verified
claim: The claim text
score: Truthfulness score (0-5, where 0 is completely false and 5 is completely true)
confidence: Confidence in the verification (0-1)
evidence: List of evidence supporting the verification
contradictions: List of evidence contradicting the claim
sources: List of sources used for verification
search_queries: List of search queries used for verification
explanation: Explanation of the verification result
is_recent_news: Flag indicating if the claim is about very recent events
Inherits from: BaseModel
__init__
Method
def __init__(self, /, **data: 'Any') -> 'None':
Create a new model by parsing and validating input data from keyword arguments.
Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
validated to form a valid model.
`self` is explicitly positional-only to allow `self` as a field name.
content_analysis_report
ContentAnalysisReport model for representing the final analysis results.
ContentAnalysisReport
Class
The complete analysis report for a website or content source.
Attributes:
url: URL of the analyzed content
domain: Domain name of the content source
topic: Main topic of the content
domain_expertise_required: List of expertise domains required to verify the content
overall_score: Overall truthfulness score of the content (0-5)
claims: List of factual claims extracted from the content
verification_results: List of verification results for the claims
summary: Summary of the analysis
recommendations: List of recommendations for the reader
analysis_date: Date when the analysis was performed
Inherits from: BaseModel
__init__
Method
def __init__(self, /, **data: 'Any') -> 'None':
Create a new model by parsing and validating input data from keyword arguments.
Raises [`ValidationError`][pydantic_core.ValidationError] if the input data cannot be
validated to form a valid model.
`self` is explicitly positional-only to allow `self` as a field name.
to_markdown_report
Method
def to_markdown_report(self) -> str:
Generate a markdown report from the analysis results.