By Artem Korsakov ·
In short: a business relationship map shows who worked with whom and in what capacity. It is useful if every connection has a basis, a period and a confirmation status. A simple line between logos proves neither the quality of the work, nor a recommendation, nor a relationship today.
Imagine looking for a contractor for a new project. Instead of a long general list, you see a familiar company, its confirmed project and participants with clear roles. A concrete next step appears: study similar work or ask for an introduction through a mutual acquaintance. This is a hypothesis of a useful scenario. It does not promise automatic sales.
Which connections are worth showing at all
Cooperation comes in different forms. A company might have been a client, contractor, supplier, partner or participant in the same event. These relationships must not be mixed: a joint presentation does not turn two organisations into business partners.
For the first version, a few connection types with a precise meaning are enough. For example, “performed work on the project”, “supplied materials”, “confirmed participation”. Each record must have dates, the party that added it, and the document or comment it relies on.
Why data provenance matters more than a pretty graph
The W3C PROV documents describe data provenance through related entities, activities and agents. This makes it possible to trace how a record came about. Source: W3C PROV-Overview.
Applying this idea to a business network is our design approach. The standard itself does not certify cooperation or guarantee a company's quality. But the principle is useful immediately: the buyer sees where a statement came from and what exactly is confirmed.
What questions the map should answer
Start with the user's specific task. A procurement manager may need to find contractors with experience on similar sites. A partner needs to understand who can introduce them to the right company. A contractor wants to show their role in the chain of work.
For each scenario, the path will be short:
- find a company or project;
- reveal the connection type and its basis;
- look at the relevant role;
- check the date and confirmation;
- choose an acceptable way to get in touch.
If a person has to peer at a hundred intersecting lines, the product has not solved their task. Next to the map, you need an ordinary list with search and filters.
What cannot be inferred from companies being neighbours
A shared counterparty does not prove that two organisations know each other personally. Working on the same project does not confirm close cooperation. Many connections do not yet mean reliability. A company without a public history does not become suspicious either.
It is not a good idea to automatically build a “trust rating” from the number of lines. That easily rewards mass addition of weak contacts instead of useful confirmations. First check whether a specific record helps make a meaningful choice.
Outdated connections also need attention. A project may have ended many years ago, the contact person may have left, and conditions may have changed. That is why history is kept as history, and current relationships are marked separately.
How to build in access control
A public map can show a general project description, with details opened on request. For example, the client's name has been approved, but the amounts and working correspondence stay within the team. Permission for one card does not automatically extend to the whole archive.
You need owner, editor, confirmer and guest roles. A guest is shown only the permitted fields. Revoking access must apply to the list, the map and exports alike. The legal grounds for disclosing materials are checked separately for the specific data and agreements.
Where AI helps and where to start
AI can suggest merging duplicate companies, identify presumed roles from a project description and find connections in response to a user's question. Suggestions are saved as drafts. A matching name or a similar address does not give the right to merge organisations automatically.
In the business network prototype, it is enough to test a few companies and one completed project. First make sure participants understand the roles the same way and agree with how they are displayed. Only then does it make sense to grow the map and the automation.
Related reading: Launching a business network: how to start before you have a large user base; Business transparency: why disclose clients, connections and terms.
Частые вопросы ### Can a network be built only from public mentions?
You can research open sources, but a mention you find must be marked as unconfirmed and stored with a link and a date. Publication, reuse and the availability of specific data require a separate check; the fact that a page is public does not automatically make any use of it permissible.
Is a complex graph database needed at the start?
Not necessarily. For a small first version, structured companies, projects and connections in an ordinary database are enough. The technology is chosen once the data volume and user queries are clear. A picture with lines is not a reason by itself.