startthat.work home

IS-06 Source dependence and copy detection (source independence analysis)

Estimates whether sources copy from each other so that their agreement is not counted as independent support and produces a dependence estimate for source pairs or groups.

Integrate across sourcesEvidence: DocumentedPreparation technique, runs in step 5.2Strongest statement: describe

What it does

Estimates whether sources copy from each other, so that their agreement is not counted as independent support. Artefact: a dependence estimate for source pairs or groups. Read which sources should be grouped as one.

When to use it

Use when several sources agree and you need to know whether they are independent or share one upstream origin.

When it does not fit

Fits when copying leaves shared errors. Misfit: a source copies from a good source, so there are no shared mistakes to detect. Co-copying and chains are harder to detect.

Result format

The result is a dependence estimate for each pair or group of sources, stating how likely it is that one copies from the other. It ends with the sources that should be grouped and counted as one.

Requirements on the material

PropertyRequirement typeRequirement
StructureApplies in some casesrequires=structured; requires=semi-structured
Content typeAlways appliesrequires=tabular; requires=text
Unit of analysisAlways appliesrequires=source
Level of detail availableAlways appliesrequires=aggregate tables
Number of sourcesAlways appliesrequires=two or more
Producer interestChanges the orderprefers=unknown
Source dependenceChanges the orderprefers=copied or derived; prefers=unknown
OriginChanges the orderprefers=secondary; prefers=tertiary; prefers=unknown
Appraises source credibilityApplies in some casesyes

Not restricted on: Variable types, Time structure, How the sample was chosen, Whether sources can be joined, Amount of material, Known data problems, Document genre, Language, Personal information, Security classification, Licence.

Minimum materialat least three sources with overlapping claims
Additional sources needednone
Not for (rules)none
Not for (text)A source copies from a good source, so no shared mistakes exist to detect. Co-copying and chains are harder to detect.

Choosing and ordering

Kind of techniquePreparation technique, runs in step 5.2
Phase tagverify and prepare
Other phase tagsnone
Kind of findingpattern; risk
Question answeredwho; what
Strongest statement it supportsdescribe
Needs as input
Makes as output
Must exist firstNone
Loop targetproceed with limit stated

Library checks by stakes

Each stakes level includes the checks of the level below it. Checks that need a person apply at high stakes only, so the process leaves them out at lower stakes (decision 38). Repeatability: deterministic.

StakesCheckPerson involvedCounts as a different method
exploratoryinput validationnoneno
decision-informinginput validationnoneno
decision-informingspot checkoptionalno
high-stakesinput validationnoneno
high-stakesspot checkoptionalno
high-stakesindependent reviewneededno
high-stakesprovenance lognoneno

Failure modes and limits

Model suitabilitymodel plus code
Typical model failureConcludes sources are independent because the wording differs, without checking shared errors.
Error sourcesinput data; statistical assumptions
Certainty limitsCertainty drops when copying leaves no shared errors, and when sources are few (imprecision).

Evidence and tag basis

Documented. In the tutorial example, two sources copy from a third, so treating them as independent would wrongly confirm everything the third provides. (Dong and Naumann, VLDB tutorial; Li et al., A survey on truth discovery, arXiv 1505.02463)

Tag basis: input profile inference; selection inference; sequencing inference; robustness inference. All tag values are inference unless a source is named.