Scoring Properties

Overview

Merge with Staging uses scoring properties to rank matched wells, well samples and non-well samples in terms of similarity to help users select the appropriate match if more than one are present.

Version: 2.0+ (Oct 2021)

Usage: Data upload, data merge and de-duplication.

Scoring properties:

Scoring properties are used by the Merge with Staging to rank similar wells or samples. When a matching well or sample is found using the matching properties it is given a score by Metis. The matching wells or samples are then ranked via their respective score with a more similar match scored higher.

In the tables below we list the scoring properties, how the scoring is calculated, and what weight is given to each component in the composite score.

The scoring properties for wells are:

Scoring property	Scoring function	Scoring weight
Lat.Well, Long.Well	Distance in meters on WGS84 great arc, quadratic decay to 150 m	5
Well Name.Well	Edit distance string similarity score	5
UWI.Well	Edit distance string similarity score	2
IHS ID.Well	Edit distance string similarity score	2
Internal#.Well	Edit distance string similarity score	2
Country.Well	Edit distance string similarity score	1

The edit distance is the number of characters that need to be changed to transform one string to another. This is scaled by the number of characters in the shorter of the two strings, to give a scaled edit distance. This is then inverted, so an edit distance of zero (a perfect match) gives a score of 100% for that criterion, while for a string of length N characters, and edit distance of N (all characters need to change) gives a score of 0%.

The scoring properties for well samples are:

Scoring property	Scoring function	Scoring weight
Name.Sample	Edit distance string similarity score	5
Base MD.Sample	Depth distance within 0.1 m, quadratic decay	2
Top MD.Sample	Depth distance within 0.1 m, quadratic decay	2
Sam date.Sample	Time separation, quadratic decline within 1 hour to zero	2
Repeat.Sample	Edit distance string similarity score	2
Internal#.Sample	Edit distance string similarity score	1
Vendor#.Sample	Edit distance string similarity score	1
Test#.Sample	Edit distance string similarity score	1
Category.Sample	Edit distance string similarity score	1
Rock type.Sample	Edit distance string similarity score	1
Liquid type.Sample	Edit distance string similarity score	1
Gas type.Gas	Edit distance string similarity score	1
Type.Sample	Edit distance string similarity score	1

The scoring properties for non-well samples are:

Scoring property	Scoring function	Scoring weight
Lat.Sample, Long.Sample	Distance in meters on WGS84 great arc, quadratic decay to 150 m	5
Name.Sample	Edit distance string similarity score	5
Sam date.Sample	Time separation, quadratic decline within 1 hour to zero	2
Repeat.Sample	Edit distance string similarity score	2
Internal#.Sample	Edit distance string similarity score	2
Vendor#.Sample	Edit distance string similarity score	1
Test#.Sample	Edit distance string similarity score	1
Category.Sample	Edit distance string similarity score	1
Rock type.Sample	Edit distance string similarity score	1
Liquid type.Sample	Edit distance string similarity score	1
Gas type.Gas	Edit distance string similarity score	1
Type.Sample	Edit distance string similarity score	1
Site type.Sample	Edit distance string similarity score	1
Country.Sample	Edit distance string similarity score	1
Field.Sample	Edit distance string similarity score	1
Location.Sample	Edit distance string similarity score	1

The scores are only used to rank the Staging wells and samples as to their degree of match. The scores are reported at a percentage, with 100% meaning that all scoring properties present in both the Staging and project well / sample are identical. In general the scoring value should not be over-interpreted - having a higher score does suggest the match is better, but the user should explore the data to confirm this.