Interest Islands

Interest Islands are RSSMonster’s personal semantic layer. They represent areas that a user repeatedly engages with—or explicitly avoids—rather than the news itself.

Examples might include artificial intelligence, photography, self-hosting, electric vehicles, or a particular game franchise. Every user’s Islands are private to that account and are learned only from that user’s articles and behavior.

An Interest Island answers what does this user consistently care about? It is not an Event, a Topic, a feed category, or simply a folder of similar articles.

Where Islands Fit

Articles and reading behavior
        |
        v
Behavioral article profiles
        |
        v
Candidate Interest Islands
        |
        v
Persist or update Islands
        |
        v
Enrich them with Topics
        |
        v
Score unread articles

Events describe individual occurrences. Topics connect recurring subjects. Interest Islands consume those semantic layers together with direct behavior, but never redefine them.

Behavioral Evidence

Island calibration starts with canonical, vectorized articles that carry an explicit behavioral signal. The current signal weights are:

Signal	Contribution
Positive feedback	`+4`
Bookmark or favorite	`+4`
Outbound click	`+1.5`, up to three clicks per article
Deep read	`+3` when attention bucket is at least three
Negative feedback	`-4`

Positive signals are reduced gradually as articles age. The default half-life is 1,460 days, with a minimum recency multiplier of 0.2, so older explicit behavior remains useful without carrying its original strength forever.

An article can contain several signals. For example, bookmarking and deeply reading an article provides stronger evidence than opening it once. Negative feedback produces signed evidence, allowing RSSMonster to learn that a semantic area should reduce rather than increase personalized relevance.

Bookmarks are especially clear evidence because they represent an explicit decision to retain an article. See Bookmarks.

Forming Candidate Islands

RSSMonster groups behavioral article profiles by cosine similarity. The strongest positive or negative evidence is processed first. An article joins the closest existing candidate when similarity is at least 0.64 by default. Otherwise, it starts another candidate until the per-user maximum—ten Islands by default—is reached. After that, remaining evidence is assigned to its nearest candidate.

Each candidate receives:

an aggregate vector weighted by the magnitude of its article evidence;
a signed weight based on average behavior plus a small breadth bonus;
counts of the signals supporting it; and
the source articles that explain why it was formed.

The weight is bounded from -1.0 to 1.0. Positive weights can boost related content, while negative weights can penalize it. A weight is a preference signal, not a quality or truthfulness rating.

The current implementation can create a candidate from one sufficiently strong vectorized article. Further related behavior makes the Island broader and more stable over subsequent calibrations.

Preserving Existing Islands

Candidate profiles are compared with the user’s stored Islands. A similarity of at least 0.78 reuses an existing Island. Its vector is blended with the new profile using a default new-evidence weight of 0.35; it is not replaced abruptly.

If no Island qualifies, RSSMonster creates a new one. This preference for updating existing Islands gives them continuity as reading habits evolve.

An unmatched Island can be archived when both conditions hold:

its average Topic-membership confidence is below 0.12; and
it has not been updated for at least 45 days.

Archived Islands remain available for inspection but are excluded from active article matching and interest scoring. A later matching profile can reactivate an archived Island.

Topic Enrichment

After behavior-derived Islands are persisted, RSSMonster evaluates stored Topics. Topics help an Island expand beyond its original source articles while remaining connected to explainable semantic subjects.

Topic profiles combine:

positive and negative behavior on Topic articles;
the Topic’s stored affinity;
bounded Event-count evidence;
overlap in the engaged articles; and
similarity in the time periods when engagement occurred.

Behavioral affinity, rather than vector similarity alone, groups Topics into candidate communities. Existing Islands are then enriched with Topics whose vectors reach a similarity of 0.62 and whose evidence-adjusted confidence reaches 0.10 by default.

Topics enrich existing behavior-derived Islands; they do not independently create the initial personal-interest layer.

Evolving Topic Membership

IslandTopic is the durable relationship between an Island and a Topic. It stores both semantic similarity and evidence-adjusted confidence.

When evidence is observed again, old and new membership values are blended. The default blend gives new evidence a weight of 0.65. Memberships not observed in the latest calibration decay to 82% of their previous confidence instead of disappearing immediately. They are removed only after confidence falls below 0.05.

This gradual evolution prevents a small change in reading behavior from reorganizing the user’s long-term interests all at once.

Island Names

RSSMonster first tries to label an Island using the nearest active semantic taxonomy name. When no taxonomy label is available, it uses the strongest related Topic names or, for article-only evidence, the strongest source article title.

Names are also disambiguated. Semantically near-identical Islands with the same normalized name can be archived as duplicates. Distinct Islands that happen to receive the same broad name are given a distinguishing phrase or suffix rather than being merged solely because their labels match.

Population Audit

Each Island keeps a bounded audit history explaining how it was populated. An audit entry can include:

contributing Topic and article IDs;
counts of related, bookmarked, clicked, and negatively rated articles; and
compact snapshots of source-article evidence.

By default, RSSMonster retains the latest 30 calibration entries and up to 300 article IDs per entry. The audit supports the Why this island exists view; it is never fed back into clustering as new semantic evidence.

Article Interest Scores

After calibration, RSSMonster recalculates interest scores for canonical, unfiltered, unread articles.

The preferred scoring path is:

Article -> Topic -> Interest Island weight

If an article belongs to several active Islands through its Topics, RSSMonster uses the Island weight with the greatest absolute magnitude. This preserves a strong negative preference as well as a strong positive one.

When an article has no applicable Topic path, RSSMonster can compare its vector directly with active Island vectors. The fallback requires similarity of at least 0.62 by default and calculates:

interest score = Island weight x vector similarity

The vector fallback replaces an existing score only when its absolute strength is greater. Topic-based scoring is preferred because it is more stable and easier to explain.

Interest scores influence sort:recommended, where positive scores boost and negative scores penalize an article. They also support Daily Briefing eligibility and semantic filtering. The score is derived output: it does not become new behavioral evidence and does not itself change an Island.

Use island:true in Search or a Smart Folder to select articles whose Event has a primary or secondary Topic linked to an active Island. island:false selects articles without such a relationship. See Search and Smart Folders.

Inspecting Your Islands

Open Settings > Islands for a read-only explanation of what RSSMonster has learned. The overview shows:

the number of active Interest Islands;
articles connected to active Islands through Topics;
articles outside Islands and overall library coverage;
each Island’s signed interest weight and active or archived state;
the behavioral source articles explaining why it exists; and
linked Topics and recently related articles.

Interest Island insights in the Settings menu

The overview is a snapshot. Use its Refresh button to fetch the latest state. Low coverage is not automatically a problem: Islands are based on explicit behavior and conservative semantic relationships, so most of a large library may remain outside them.

Calibration and Normal Crawls

A normal crawl does not rebuild the user’s Islands. It assigns Events and Topics to new articles, then scores those new unread articles against the existing active Islands. This keeps routine crawling bounded.

To recalibrate Islands for every user and then refresh article interest scores, run from the server directory:

npm run islands

The historical semantic pipeline also recalibrates Islands after rebuilding Events and Topics:

npm run semantic:all

These operations inspect behavioral history and update semantic relationships, so run them deliberately on large multi-user libraries.

Advanced Server Tuning

Most installations should use the defaults. The main controls are:

Variable	Default	Effect
`MAX_INTEREST_ISLANDS`	`10`	Maximum behavioral communities formed for each user.
`ISLAND_ARTICLE_AFFINITY_THRESHOLD`	`0.64`	Similarity needed to group behavioral articles into one candidate.
`ISLAND_ARTICLE_SIGNAL_THRESHOLD`	`0.05`	Minimum absolute behavioral score admitted to article clustering.
`ISLAND_PROFILE_MATCH_THRESHOLD`	`0.78`	Similarity needed to update an existing Island instead of creating one.
`ISLAND_VECTOR_ALPHA`	`0.35`	Weight of new profile evidence when updating an Island vector.
`ISLAND_RECENCY_HALF_LIFE_DAYS`	`1460`	Half-life for positive behavioral evidence.
`ISLAND_RECENCY_MIN_WEIGHT`	`0.2`	Minimum retained multiplier for old positive behavior.
`ISLAND_TOPIC_ENRICHMENT_SIMILARITY_THRESHOLD`	`0.62`	Semantic similarity needed for Topic enrichment.
`ISLAND_TOPIC_CONFIDENCE_THRESHOLD`	`0.10`	Minimum evidence-adjusted Topic membership confidence.
`ISLAND_MEMBERSHIP_BLEND`	`0.65`	New-evidence share when refreshing Topic memberships.
`ISLAND_MEMBERSHIP_DECAY`	`0.82`	Confidence retained for an unobserved membership.
`ISLAND_MEMBERSHIP_MIN_CONFIDENCE`	`0.05`	Membership confidence below which a link is removed.
`ISLAND_ARTICLE_SCORE_THRESHOLD`	`0.62`	Similarity required by direct article-vector fallback scoring.
`ISLAND_ARCHIVE_CONFIDENCE_THRESHOLD`	`0.12`	Low-confidence condition for archiving an inactive Island.
`ISLAND_ARCHIVE_STALE_DAYS`	`45`	Minimum inactive age before low-confidence archival.
`ISLAND_DUPLICATE_NAME_SIMILARITY_THRESHOLD`	`0.92`	Similarity at which same-name Islands are treated as duplicates.
`ISLAND_AUDIT_MAX_RUNS`	`30`	Maximum retained population-audit entries.
`ISLAND_AUDIT_MAX_ARTICLE_IDS`	`300`	Maximum stored article IDs per audit entry.

Additional Topic-community controls include ISLAND_TOPIC_AFFINITY_THRESHOLD (0.12), ISLAND_MAX_COMMUNITIES_PER_TOPIC (2), ISLAND_ENGAGEMENT_TIME_BUCKET_HOURS (12), and ISLAND_TEMPORAL_AFFINITY_WEIGHT (0.65).

Island thresholds interact: permissive settings can combine unrelated interests, while strict settings can create fragmented or sparsely connected Islands. ISLAND_DEBUG=true enables detailed calibration, membership, and scoring diagnostics. EVENT_DEBUG=true also enables Island debug output.