Home→ Graphs

Semantic relationship layer

Graphs

Turning Brand Mentions into visible relationship patterns.

A Brand Mention count tells us that a brand appeared. A graph asks a different question: what was the brand connected to when it appeared?

Explore graph library →View source dataset →

BRAND
CONTEXT
RELATIONSHIPSObserved textual relationships
not verified causality

536relationship records

276unique Brand → Relation → Entity pairs

139brands with detected semantic relationships

98entities or relationship terms

01FROM COUNT TO CONTEXT

The semantic calibration behind the graph layer contains:

536 relationship records
276 unique Brand → Relation → Entity pairs
139 brands with at least one detected semantic relationship
98 entities or relationship terms

Those records make it possible to move from a flat list of mentions to a visible network of products, problems, features, use cases, results, locations and co-mentioned brands.

The graphs do not add new evidence.

They reorganize the semantic relationships already extracted from the analysed text so that recurring patterns become easier to see.

Graph library

Five views of the same semantic layer.

Each graph starts with a different analytical question. Together they show how Brand Mentions connect to products, problems, use cases, outcomes, places and other brands.

FULL SEMANTIC VIEW

Brand Context Graph

What it answers

What surrounds a brand across many mentions?

Brand Context Graph combines the semantic relationships detected around a brand into one view.

A brand can be connected with:

products, services, features, problems, use cases, results, locations, and other brands.

The value of the graph is not that it makes the brand look more complex.

The value is that it shows which associations recur and which remain isolated.

Across many mentions, those recurring relationships begin to describe the brand's observed semantic territory.

WHAT IT DOES NOT PROVE

The graph does not prove that every relationship is a verified external fact.

It does not prove that a Result relationship is causal.

It does not prove that co-mentioned brands are direct competitors.

It visualizes observed textual relationships.

Open Brand Context Graph →

ASSOCIATION BREADTH

Semantic Territory

What it answers

How broad or narrow is the set of meanings associated with a brand?

Some brands may appear mainly around one product or one use case.

Others may be connected with several problems, services, features, products and outcomes.

Semantic Territory is a way to compare those patterns.

A narrow territory can indicate focus.

A broader territory can indicate wider contextual coverage.

Neither is automatically better.

The graph is useful when comparing the language surrounding a brand with the positioning the company intends to create.

Why this matters

A company may describe itself one way on its own website while third-party discussions connect it with a different set of products, problems or use cases.

The graph helps make that difference visible.

Open Semantic Territory →

NEED-DRIVEN PATH

Problem to Brand

What it answers

Which brands appear when a discussion starts with a problem or need?

This graph reverses the usual direction of brand analysis.

Instead of starting with a company name, it starts with a problem.

The semantic calibration includes:

23 Problem relationship records
288 Use Case records
23 Result relationship records
64 Competitor / Co-Mention records

Those relationships make it possible to inspect need-driven paths where brands appear after the initial problem is described.

WHAT IT DOES NOT PROVE

A Problem relationship does not prove that the brand solved the problem.

A Result relationship does not prove causality.

Equal Problem and Result counts do not mean that they form matched one-to-one paths.

The graph shows structure, not verified outcomes.

Open Problem to Brand →

PRODUCT ASSOCIATIONS

Brand to Product

What it answers

Which products or product categories are associated with a brand in the analysed discussion context?

The semantic calibration contains 48 Product relationships.

A Brand → Product graph makes those associations visible.

This can show whether a brand appears repeatedly around one product category or across several different product contexts.

It can also reveal overlap between brands that appear around similar products.

WHAT IT DOES NOT PROVE

The graph is not an official product catalog.

It does not prove sales volume.

It does not prove market share.

It records products that were associated with the brand in the analysed text.

Open Brand to Product →

BRAND NEIGHBORHOODS

Brand Co-Mention Network

What it answers

Which brands appear together inside the same relevant context?

The calibration contains 64 Competitor / Co-Mention records.

A co-mention network can reveal observed brand neighborhoods.

Two brands may appear together because users compare them.

They may be alternatives.

They may be complementary.

They may simply occupy the same category context.

Why this graph needs a strong caveat

Co-mention is not the same thing as competition.

A network edge records proximity in the analysed text.

It does not automatically tell us why the brands appeared together.

That reason has to be interpreted from the surrounding context.

Open Brand Co-Mention Network →

Interpretation

Graphs are a view of evidence, not evidence by themselves.

Why graphs add value to the research

The graph layer is useful because the same relationship can recur across many mentions.

A product can appear again and again around the same brand.

A use case can connect several brands.

A problem can lead to different alternatives.

A co-mention pattern can reveal a recurring consideration set.

Those structures are difficult to see in a spreadsheet.

A graph makes them visible.

But the visual layer remains subordinate to the source data.

The graph is a way to inspect the relationships.

It is not a replacement for the underlying evidence.

Frequency and diversity remain separate

The semantic calibration contains 536 relationship records but 276 unique pairs.

That difference is important.

The total relationship count shows how often relationships were recorded.

The unique-pair count shows how many distinct Brand → Relation → Entity combinations exist after normalization.

A graph can display either frequency, diversity or both.

The reader should know which one is being shown.

Otherwise repeated relationships can be mistaken for broader semantic territory.

RESEARCH BOUNDARY

What the graph layer does not claim

The graphs do not claim that semantic breadth equals authority.

They do not claim that more relationships automatically create more value.

They do not claim that co-mention means competition.

They do not claim that Result means causality.

They do not claim that the observed relationships automatically improve AI visibility.

The graph layer has a narrower purpose:

to make the context surrounding Brand Mentions visible and measurable.

Where to start

Choose the graph by the question you need to answer.

01Start with Brand Context Graph for the full semantic view.→ 02Use Semantic Territory to inspect breadth and concentration of associations.→ 03Use Problem to Brand when the starting point is a need rather than a company name.→ 04Use Brand to Product to inspect product associations.→ 05Use Brand Co-Mention Network to inspect which brands appear together in relevant context.→