Truth Ranges

Truth ranges refer to the highest and lowest values identified for a specific piece of data, such as numeric values like pH levels, temperature, or speed, etc. These ranges are created when there are multiple sources providing different values for the same entity-attribute-value (E-A-V) structure, which represents a piece of information (like a substance and its pH level).

🎯 Purpose

Truth ranges help search engines or systems deal with inconsistencies across different sources. Instead of selecting just one value, the system identifies the range of values provided by different sources and can present the most accurate and comprehensive answer.

🤔 How it Works?

If there are 90 web sources that mention the specific alchemy formula, and only 37 of them involve something related to pH level, the search engine has to filter the sources based on their attribute involvement. 37 different web sources for pH level might give 37 different pH level suggestions. In these types of situations, search engines can use their own Knowledge Bases that are fed with Data Commons, Freebase, or manual Knowledge Base improvements, or they can create a truth range with various corners.

When the data from different sources does not align, truth ranges come into play:

🔍 Collecting Values: The search engine gathers all possible values for a particular query. For example, for the pH level of a substance, various websites may provide different values.
📏 Finding the Range: After collecting values, the system identifies the minimum and maximum points. This process allows the creation of a truth range that encompasses all the possible data points.
🚀 Boosting Rankings: If no single consensus is available, a source that provides a range of values (e.g., "The pH level of X varies between Y and Z") can be favored in rankings, as it helps users understand the variability in the data.

🔬 Example: pH Levels of a Substance

Suppose there are 37 sources giving different pH levels for a chemical:

One source says the pH level is between 5 and 6.
Another source claims it's between 4 and 5.
A third source suggests it's between 6 and 7.

In this case, instead of choosing one source, the search engine identifies the truth range from 4 to 7 as the full spectrum of possible values. By using this range, the search engine provides a more accurate representation of the possible pH levels, even if no single source agrees on the exact value.

🔥 Truth Ranges Example: Melting Point of a Metal

Imagine you're researching the melting point of a specific metal, such as aluminum, and you come across several different sources that provide slightly varying information:

Source A: states that the melting point of aluminum is between 659°C and 660°C.
Source B: claims the melting point is between 660°C and 661°C.
Source C: mentions a broader range, stating the melting point of aluminum is between 657°C and 662°C.