At Vurvey, we're committed to providing you with the most accurate and relevant information possible. Our AI-powered platform doesn't just give you answers – it shows you why those answers are relevant by grounding them in real data from our extensive surveys and campaigns.
This approach ensures transparency and builds trust, allowing you to see firsthand how our AI reaches its conclusions. This article will guide you through the process, breaking down the technology behind it into easy-to-understand concepts.
What is Grounding, and Why Does It Matter?
Imagine asking a question and receiving an answer without any context or source. You might wonder, "How do they know that?" or "Where did they get that information?" This is where grounding comes in.
Grounding connects an AI's response directly to the data it used to formulate that response. It's like providing footnotes or citations but in a much more intuitive and dynamic way. Instead of simply telling you something, Vurvey shows you the evidence, increasing transparency and giving you confidence in the information presented.
Vurvey's 3-Step Approach to Grounding
Vurvey's grounding process ensures that the information you receive is accurate and directly relevant to your query. We achieve this through a three-step algorithm:
Context Relevancy: Finding the Right Information
Think of this step as a highly efficient librarian searching a vast library. When you ask a question, Vurvey's AI doesn't just randomly scan through all the data. Instead, it uses sophisticated algorithms to identify the most relevant information related to your query.
Keywords and Concepts: Vurvey analyzes your question to identify key terms and concepts. For example, if you ask, "What are the main concerns of young adults today?" Vurvey will focus on data related to "young adults," "concerns," "worries," "challenges," and similar terms.
Dataset Matching: Vurvey then searches through its extensive survey data collection, identifying datasets that align with the identified keywords and concepts. This ensures that the AI is looking at the most pertinent information.
Context Precision: Ranking the Best Evidence
Once Vurvey has identified potentially relevant datasets, the next step is prioritizing them. Not all data is created equal – some datasets will be more closely aligned with your query than others.
Relevance Scoring: Vurvey employs machine learning algorithms to analyze and score each dataset based on its relevance to your question. Datasets that closely match the identified keywords and concepts and contain statistically significant responses related to your query will receive higher scores.
Prioritizing the Best: This ranking process ensures that the most relevant and informative datasets are given priority when generating a response.
Answer Relevancy: Connecting the Dots
The final step is where Vurvey's AI crafts a response, drawing directly from the ranked datasets. This isn't just about pulling out random quotes or data points – it's about synthesizing the information into a clear, concise, and insightful answer.
Evidence-Based Responses: Vurvey's AI uses the highest-ranked datasets to respond directly to your question. Instead of giving you a generic answer, Vurvey provides information grounded in real data.
Grounding Snippets: To ensure transparency, Vurvey highlights the specific data points, survey responses, or campaign results that support the provided answer. These grounding snippets are presented alongside the response, allowing you to see exactly where the information came from.
The Power of a Weighted Score
Vurvey combines the three metrics – context relevancy, context precision, and answer relevancy into a single, weighted score to ensure the most accurate and relevant grounding. This means that each step carries a different level of importance in the final analysis.
For example, finding highly relevant datasets (context relevancy) might be more important than simply identifying many loosely related datasets. Similarly, providing a concise answer directly supported by the top-ranked datasets (answer relevancy) will carry more weight than an answer that only partially reflects the data.
Grounding: Transparency You Can Trust
Vurvey's grounding system goes beyond traditional methods of providing information. By grounding its responses in real data and showing you the evidence, Vurvey empowers you to understand:
Let's say a user asks Vurvey, "What are the biggest challenges facing small business owners in the current economy?"
Here's how the grounding system might work:
Context Relevancy: Vurvey identifies the key terms "challenges," "small business owners," and "current economy." It then searches for datasets that contain these terms or closely related concepts. It might prioritize datasets from surveys specifically targeting small business owners or those focused on economic conditions.
Context Precision: Vurvey analyzes the identified datasets and ranks them based on relevance. A dataset with survey responses from hundreds of small business owners about their current challenges would likely rank highly. While still relevant, a dataset about general economic trends might be ranked lower.
Answer Relevancy: Vurvey's AI then crafts a response. Instead of saying something generic like, "Small business owners are facing many challenges," it might say:
"According to a recent Vurvey survey of 500 small business owners, the top three challenges they face are inflation (cited by 72% of respondents), supply chain disruptions (61%), and difficulty hiring qualified employees (54%)."
Alongside this response, Vurvey would provide grounding snippets. These might include:
A graph from the survey data shows the percentage of respondents who cited each challenge.
Direct quotes from small business owners explain how these challenges impact their businesses.
By grounding its response in this way, Vurvey provides:
Credibility: The answer is clearly backed up by real data.
Transparency: Users can see exactly where the information came from.
Actionability: The specific data points (percentages, quotes) give users a clearer understanding of the challenges and their potential impact.
We are very early in developing our groundings feature. While you can not pinpoint specific videos with consumers verbatim with the links at this time, this will be added in the future as we continue to improve and refine our algorithm through user use and feedback.