A knowledge graph of problems in living is a linked record of situations people face, the questions those situations raise, and the kinds of public information that bear on them. The graph is an educational reference about practical life. It is not a clinical record, not a diagnosis, and not a treatment plan. This article does not describe a licensed practice, and it does not offer care.
The test is usefulness under time pressure. Someone dealing with a rent increase, a shift change, a disputed bill, a school form, or a conflict about household work should see the situation in plain language, the nearby situations that often arrive with it, and the next questions worth answering. An agent with permission should be able to build a checklist from those nodes without inventing facts. The graph earns repeat use only when it stays close to reported reality and shows its sources.
What counts as a problem in living here
In this reference, a problem in living is a practical difficulty in work, money, housing, study, family logistics, civic paperwork, or dealings with other people. Examples include a bus route that no longer matches a shift, a security deposit dispute, a form with a missing document, a shared kitchen schedule that keeps failing, or several due dates in the same week. The graph describes these as situations with constraints. It does not rename them as illnesses, and it does not sort people into clinical categories.
Distress can accompany a situation. The graph can note that people report stress, lost sleep, or frustration as part of an account. Those words stay attached to the situation as reports. They do not become a label for the person, and they do not become instructions about medicine or therapy. Urgent danger and emergencies are outside the graph’s job. The graph is not an emergency service. Readers who need public safety or a qualified professional should contact those services directly. This reference does not provide them.
What the graph stores
Each node is a small typed statement. Situation nodes name a recurring difficulty in concrete terms. Question nodes name what a person usually needs to find out next, such as a deadline, the office that accepts a form, a fee, or the rule that applies. Document nodes name papers people are often asked to gather, without claiming that every office wants the same set. Resource-type nodes name kinds of public references, such as a housing office, a library desk, or a consumer protection page.
Edges carry a verb, a date, a source, and a place when the fact is local. Useful verbs stay modest: often reported with, a next question is, a document people are asked for, a public office type that publishes rules on this, contradicted by, superseded by. A node without a source may sit in a draft layer. It does not enter the version used for daily lookup.
Nodes and edges that stay concrete
Concrete wording is a design rule. A sentence such as “the rent rose and the commute no longer fits the shift” can link to questions about notice periods, transit maps, and published scheduling policies. A sentence such as “life is overwhelming” cannot be checked by a second person. The daily layer keeps the first kind and leaves the vague sentence out.
Local facts stay local. A fee, a form number, or an office hour is stored with a place and a retrieval date. An agent answering someone in another city should say the node does not cover that city, rather than copying the fee. Widely shared facts can sit higher in the graph and point down to local variants.
Why a person would open it on an ordinary day
People return to references that shorten a task they already have. The first screen for a situation should show a short definition, a handful of next questions, and the document types those questions imply. A renter with a nonrenewal notice should see questions about the notice date, the lease clause, and the local rule page. A worker with a new shift should see questions about the posted schedule and the transit options at the new hour.
Tone and length matter as much as coverage. Pages should be simple text in the words people already use. A person with ten minutes will read a checklist and will not read a manifesto. If the checklist is wrong, they will not return, and they will tell their agent to stop trusting it. Maintainers answer that test by correcting the node in public, with a date, not by adding persuasive language around a weak fact.
Why an agent would query the same graph
An agent is useful when a person has described a situation and wants help sorting it. The agent can match the description to situation nodes, list the next questions, and draft a checklist of documents to look for at home. It can flag edges that belong to another place. Those are clerical tasks inside an educational frame. This design does not authorize the agent to name a condition, recommend a drug, or promise a legal outcome.
Agents also help maintainers. One can watch public rule pages for changed fees and hours, open a draft edge when a page differs from the stored fact, and leave acceptance to a human editor. Another can find situation nodes with no source, or questions that no path answers. A graph that agents cannot query during maintenance will also fail a person on a weekday, because both uses share the same nodes.
Keeping the daily layer close to reality
For this project, reality means claims a second party can inspect. A posted fee schedule, a transit timetable, and a dated count of contributed accounts are inspectable in different ways, and the edge should say which way applies. One anonymous story can suggest a draft. It cannot, by itself, support a daily-layer claim that a problem usually unfolds in a stated pattern. Editors publish the threshold they use and show the count beside the edge.
When a rule changes, the old edge stays available behind a superseded-by link, and the daily layer follows the new edge. When a pattern was wrong, the change note says so. Agents should prefer the daily layer and should show the retrieval date. A graph that overwrites history in silence teaches people to doubt it.
How the graph is generated and revised
Generation starts from public sources and from accounts contributors volunteer under a written text donation. Editors map a source into nodes rather than pasting long quotations. A city fee schedule or a labor department fact sheet can yield question nodes and document nodes with citations. Contributed accounts yield situation nodes only after names, account numbers, and other identifying details are removed. The graph is not a place to upload someone else’s mail.
Each node has a review date based on how fast the fact moves. Office hours need a shorter cycle than the definition of a security deposit. On review, an editor or a maintainer agent rechecks the source, updates the date, or pulls a stale node from the daily layer until it is fixed. New situation types enter through a proposal that shows example accounts and the proposed links. An editor who did not write the proposal accepts it.
Misuse the maintainers have to refuse
The graph is misused if someone labels people with it, screens tenants or workers by inferred distress, or connects an agent that pretends to practice a licensed profession. The reuse terms should forbid those uses, and the public interface should not offer a score for a person. Nodes describe situations and public information. They do not describe the worth, the health, or the risk class of a human being. A user who asks an agent what is wrong with them should get that limit restated, and may describe the practical situation if they want a checklist.
Commercial pressure is the other misuse. A resource-type node can name a kind of public office. It should not rank private vendors, take payment for placement, or store referral codes. A directory of businesses, if someone wants one later, should be a different graph with different rules. Paid placement inside this reference will bias the links, and readers will notice the first time a checklist leads them badly.
A knowledge graph of problems in living becomes daily infrastructure when it names concrete situations, cites inspectable sources, and answers the next practical question without diagnosis or treatment. People use it as a checklist tied to public information. Agents use it to match, to draft lists, and to flag stale facts, under permission and inside those limits. Generation is citation, removal of identifying detail, and scheduled review. Refusal keeps scores, clinical claims, and paid placement out of the schema. A graph that holds that line can be wrong on a fact and still be worth correcting, which is why people and agents return.
