5 Commits

Author SHA1 Message Date
KS Jannette
362a47f88a Upgraded embedding model to voyage-3.5, updated README.md 2026-08-01 05:35:06 -04:00
KS Jannette
c6b07ccb56 fix minor css issue on stickies 2026-08-01 04:50:40 -04:00
KS Jannette
8e79e78006 hotfix 2026-08-01 04:41:31 -04:00
19d4a82c90 Merge pull request 'Re-aligned heuristic, updated readme, added code comment' (#3) from BUG-cohesion-scoreUI-display into master
Reviewed-on: #3
2026-08-01 08:13:29 +00:00
KS Jannette
6f09b6ecdc Re-aligned heuristic, updated readme, added code comment 2026-08-01 04:06:39 -04:00
6 changed files with 193 additions and 174 deletions

View File

@@ -8,11 +8,15 @@ In kongruity, the artifacts become "sticky notes." A board full of them looks ch
With a click, they are semantically evaluated, grouped into thematic clusters with descriptive headers, rankable and exportable to project planning and execution tools. With a click, they are semantically evaluated, grouped into thematic clusters with descriptive headers, rankable and exportable to project planning and execution tools.
## Voyage AI voyage-3.5
![Embedding model benchmarking.](Voyage.jpg)
## Clustering and evaluation: methodology ## Clustering and evaluation: methodology
Two models run in parallel, and neither sees the other's work. Anthropic's `claude-sonnet-5` (`backend/services/clustering.service.js`) reads the raw text of every note and groups them into labeled thematic clusters. Two models run in parallel, and neither sees the other's work. Anthropic's `claude-sonnet-5` (`backend/services/clustering.service.js`) reads the raw text of every note and groups them into labeled thematic clusters.
At the same time, Voyage AI's voyage-3 model (`backend/services/embedding.service.js`) converts each note's text into a numeric representation of its semantic meaning aka vector. At the same time, Voyage AI's voyage-3.5 model (`backend/services/embedding.service.js`) converts each note's text into a numeric representation of its semantic meaning aka vector.
Once the LLM returns, kongruity scores that grouping (`backend/services/validation.service.js`) using an established silhouette coefficient, with cosine distance rather than Euclidean as the proximity metric. Once the LLM returns, kongruity scores that grouping (`backend/services/validation.service.js`) using an established silhouette coefficient, with cosine distance rather than Euclidean as the proximity metric.
@@ -30,14 +34,18 @@ Average silhouette width is a widely-used measure of clustering quality. Higher
2. How well-separated each cluster is from its nearest neighboring cluster. 2. How well-separated each cluster is from its nearest neighboring cluster.
Although the coefficient is mathematically bounded by [−1, 1], cosine distance between high-dimensional text embeddings is compressed: unrelated notes sit close to orthogonal, so both the within-cluster and nearest-cluster distances land near 0.8. Because silhouette divides the gap between them by the larger of the two, the practical range on embedding data is roughly [−0.05, 0.10] rather than the full interval.
The bands below are therefore calibrated against that observed range. On the seed board, the five ideal thematic clusters score 0.09; swapping a few notes between clusters drops it to 0.06; a scrambled assignment falls below zero.
The score appears above the results with a plain-language band: The score appears above the results with a plain-language band:
- **0.70 and above** — Strong - **0.07 and above** — Strong
- **0.40 to 0.69** — Moderate - **0.04 to 0.06** — Moderate
- **0.10 to 0.39** — Weak - **0.01 to 0.03** — Weak
- **Below 0.10** — Poor - **Below 0.01** — Poor
Silhouette values are archetypically bounded below 1.0 for real-world data, so the number is best read as a relative measure. See Hugo Sträng, Tai Dinh. An upper bound on the silhouette evaluation metric for clustering. Pattern Recognition, Volume 178, 2026, 113402, ISSN 0031-3203. A score near 0.00 means the grouping is no better than chance. Bands are specific to `voyage-3` cosine distance and would need recalibration behind a different embedding model. See Hugo Sträng, Tai Dinh. An upper bound on the silhouette evaluation metric for clustering. Pattern Recognition, Volume 178, 2026, 113402, ISSN 0031-3203.
## Organizing clusters, exporting to workflow software ## Organizing clusters, exporting to workflow software

BIN
Voyage.jpg Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 114 KiB

View File

@@ -27,7 +27,7 @@ export const embedNotes = async (notes) => {
for await (const chunk of batches) { for await (const chunk of batches) {
const response = await client.embed({ const response = await client.embed({
input: chunk.map((n) => n.text), input: chunk.map((n) => n.text),
model: "voyage-3", model: "voyage-3.5",
}); });
response.data.forEach((item, i) => { response.data.forEach((item, i) => {

View File

@@ -6,132 +6,136 @@ import Sticky from './sticky';
import Button from './button'; import Button from './button';
import '../styles/stickies.css'; import '../styles/stickies.css';
// In practice, silhouette on cosine distance between text embeddings occupies roughly
// [-0.05, 0.10], not strict theoretical [-1, 1]: near-orthogonal vectors put both the within- and
// nearest-cluster distances close to 0.8, and the coefficient divides their gap
// by the larger. These bands are calibrated to that range for voyage-3. see README, Reading the cohesion score
const scoreLabel = (score: number): string => { const scoreLabel = (score: number): string => {
if (score >= 0.7) return 'Strong'; if (score >= 0.07) return 'Strong';
if (score >= 0.4) return 'Moderate'; if (score >= 0.04) return 'Moderate';
if (score >= 0.1) return 'Weak'; if (score >= 0.01) return 'Weak';
return 'Poor'; return 'Poor';
}; };
const Stickies = () => { const Stickies = () => {
const { data: stickies, isLoading, error } = useGetStickies(); const { data: stickies, isLoading, error } = useGetStickies();
const { mutate: cluster, data: clusterResponse, isPending } = useClusterStickies(); const { mutate: cluster, data: clusterResponse, isPending } = useClusterStickies();
const [rankedClusters, setRankedClusters] = useState<RankedCluster[]>([]); const [rankedClusters, setRankedClusters] = useState<RankedCluster[]>([]);
const dragIndex = useRef<number | null>(null); const dragIndex = useRef<number | null>(null);
const [dragOverIndex, setDragOverIndex] = useState<number | null>(null); const [dragOverIndex, setDragOverIndex] = useState<number | null>(null);
useEffect(() => { useEffect(() => {
if (clusterResponse?.clusters) { if (clusterResponse?.clusters) {
setRankedClusters( setRankedClusters(
clusterResponse.clusters.map((c, i) => ({ ...c, rank: i + 1 })) clusterResponse.clusters.map((c, i) => ({ ...c, rank: i + 1 }))
); );
} }
}, [clusterResponse]); }, [clusterResponse]);
if (isLoading) return <div>Loading...</div>; if (isLoading) return <div>Loading...</div>;
if (error) return <div>Error: {error.message}</div>; if (error) return <div>Error: {error.message}</div>;
const score = clusterResponse?.score; const score = clusterResponse?.score;
const handleCluster = () => { const handleCluster = () => {
cluster(); cluster();
}; };
const buildStickyMap = (): Map<string, StickyType> => { const buildStickyMap = (): Map<string, StickyType> => {
const map = new Map<string, StickyType>(); const map = new Map<string, StickyType>();
stickies?.forEach((s) => map.set(s.id, s)); stickies?.forEach((s) => map.set(s.id, s));
return map; return map;
}; };
const stickyMap = buildStickyMap(); const stickyMap = buildStickyMap();
const renderStickies = (items: StickyType[]) => const renderStickies = (items: StickyType[]) =>
items?.map((sticky) => <Sticky key={sticky.id} sticky={sticky} />); items?.map((sticky) => <Sticky key={sticky.id} sticky={sticky} />);
const handleDragStart = (index: number) => { const handleDragStart = (index: number) => {
dragIndex.current = index; dragIndex.current = index;
}; };
const handleDragOver = (e: DragEvent, index: number) => { const handleDragOver = (e: DragEvent, index: number) => {
e.preventDefault(); e.preventDefault();
setDragOverIndex(index); setDragOverIndex(index);
}; };
const handleDragLeave = () => { const handleDragLeave = () => {
setDragOverIndex(null); setDragOverIndex(null);
}; };
const handleDrop = (targetIndex: number) => { const handleDrop = (targetIndex: number) => {
const sourceIndex = dragIndex.current; const sourceIndex = dragIndex.current;
if (sourceIndex === null || sourceIndex === targetIndex) { if (sourceIndex === null || sourceIndex === targetIndex) {
dragIndex.current = null; dragIndex.current = null;
setDragOverIndex(null); setDragOverIndex(null);
return; return;
} }
const reordered = [...rankedClusters]; const reordered = [...rankedClusters];
const [moved] = reordered.splice(sourceIndex, 1); const [moved] = reordered.splice(sourceIndex, 1);
reordered.splice(targetIndex, 0, moved); reordered.splice(targetIndex, 0, moved);
setRankedClusters(reordered.map((c, i) => ({ ...c, rank: i + 1 }))); setRankedClusters(reordered.map((c, i) => ({ ...c, rank: i + 1 })));
dragIndex.current = null; dragIndex.current = null;
setDragOverIndex(null); setDragOverIndex(null);
}; };
const handleDragEnd = () => { const handleDragEnd = () => {
dragIndex.current = null; dragIndex.current = null;
setDragOverIndex(null); setDragOverIndex(null);
}; };
console.log(score?.toFixed(2))
return ( return (
<div className="stickies-container"> <div className="stickies-container">
<Button onClick={handleCluster} isLoading={isPending} label="Group Stickies By Topic" /> <Button onClick={handleCluster} isLoading={isPending} label="Group Stickies By Topic" />
{rankedClusters.length > 0 ? ( {rankedClusters.length > 0 ? (
<div className="clusters-container"> <div className="clusters-container">
{score != null && ( {score != null && (
<div className="cohesion-score"> <div className="cohesion-score">
Cluster cohesion: <strong>{score.toFixed(2)}</strong> — {scoreLabel(score)} Cluster cohesion: <strong>{scoreLabel(score)}</strong>
</div> </div>
)} )}
{rankedClusters.map((group, index) => ( {rankedClusters.map((group, index) => (
<div <div
key={group.label} key={group.label}
className={`cluster-group cluster-draggable${dragOverIndex === index ? ' cluster-drag-over' : ''}`} className={`cluster-group cluster-draggable${dragOverIndex === index ? ' cluster-drag-over' : ''}`}
draggable draggable
onDragStart={() => handleDragStart(index)} onDragStart={() => handleDragStart(index)}
onDragOver={(e) => handleDragOver(e, index)} onDragOver={(e) => handleDragOver(e, index)}
onDragLeave={handleDragLeave} onDragLeave={handleDragLeave}
onDrop={() => handleDrop(index)} onDrop={() => handleDrop(index)}
onDragEnd={handleDragEnd} onDragEnd={handleDragEnd}
> >
<div className="cluster-header"> <div className="cluster-header">
<span className="cluster-rank" aria-label={`Priority ${group.rank}`}> <span className="cluster-rank" aria-label={`Priority ${group.rank}`}>
{group.rank} {group.rank}
</span> </span>
{group.rank === 1 && ( {group.rank === 1 && (
<span className="cluster-reorder-hint">Drag and drop to reorganize cluster priority</span> <span className="cluster-reorder-hint">Drag and drop to reorganize cluster priority</span>
)} )}
<h3 className="cluster-label">{group.label}</h3> <h3 className="cluster-label">{group.label}</h3>
<span className="cluster-drag-handle" aria-hidden="true">⠿</span> <span className="cluster-drag-handle" aria-hidden="true">⠿</span>
</div> </div>
<div className="stickies-grid"> <div className="stickies-grid">
{renderStickies( {renderStickies(
group?.noteIds group?.noteIds
.map((id) => stickyMap?.get(id)) .map((id) => stickyMap?.get(id))
.filter((s): s is StickyType => !!s) .filter((s): s is StickyType => !!s)
)} )}
</div> </div>
</div> </div>
))} ))}
</div>
) : (
<div className="stickies-grid">
{renderStickies(stickies ?? [])}
</div>
)}
</div> </div>
) : ( );
<div className="stickies-grid">
{renderStickies(stickies ?? [])}
</div>
)}
</div>
);
}; };
export default Stickies; export default Stickies;

View File

@@ -1,100 +1,107 @@
.stickies-grid { .stickies-grid {
display: flex; display: flex;
flex-wrap: wrap; flex-wrap: wrap;
gap: 16px; gap: 16px;
justify-content: center; justify-content: center;
margin-top: 18px; margin-top: 18px;
padding: 24px; padding: 24px;
} }
.stickies-container { .stickies-container {
margin: 36px 0px 36px 0px; margin: 36px 0px 36px 0px;
} }
.clusters-container { .clusters-container {
display: flex; display: flex;
flex-direction: column; flex-direction: column;
gap: 32px; gap: 32px;
padding: 24px 0; padding: 24px 0;
} }
.cluster-group { .cluster-group {
border: 1px solid #6dd6f4; border: 1px solid #6dd6f4;
border-radius: 8px; border-radius: 8px;
padding: 16px; padding: 16px;
} }
.cluster-draggable { .cluster-draggable {
cursor: grab; cursor: grab;
transition: box-shadow 0.2s ease, border-color 0.2s ease, transform 0.15s ease; transition: box-shadow 0.2s ease, border-color 0.2s ease, transform 0.15s ease;
} }
.cluster-draggable:active { .cluster-draggable:active {
cursor: grabbing; cursor: grabbing;
} }
.cluster-drag-over { .cluster-drag-over {
border-color: #ffb7ce; border-color: #ffb7ce;
box-shadow: 0 0 12px rgba(255, 183, 206, 0.4); box-shadow: 0 0 12px rgba(255, 183, 206, 0.4);
transform: scale(1.01); transform: scale(1.01);
} }
.cluster-header { .cluster-header {
display: flex; position: relative;
align-items: center; display: flex;
gap: 12px; align-items: center;
margin-bottom: 16px; gap: 12px;
margin-bottom: 16px;
min-height: 32px;
} }
.cluster-rank { .cluster-rank {
display: flex; display: flex;
align-items: center; align-items: center;
justify-content: center; justify-content: center;
width: 32px; width: 32px;
height: 32px; height: 32px;
border-radius: 50%; border-radius: 50%;
background: #6dd6f4; background: #6dd6f4;
color: #1a1a2e; color: #1a1a2e;
font-weight: 700; font-weight: 700;
font-size: 0.95em; font-size: 0.95em;
flex-shrink: 0; flex-shrink: 0;
} }
.cluster-reorder-hint { .cluster-reorder-hint {
font-size: 0.8em; font-size: 0.8em;
color: #9ca3af; color: #9ca3af;
font-style: italic; font-style: italic;
white-space: nowrap; white-space: nowrap;
flex-shrink: 0; flex-shrink: 0;
} }
.cluster-label { .cluster-label {
margin: 0; position: absolute;
font-size: 1.2em; left: 50%;
font-weight: 600; transform: translateX(-50%);
flex: 1; max-width: 50%;
margin: 0;
font-size: 1.2em;
font-weight: 600;
pointer-events: none;
} }
.cluster-drag-handle { .cluster-drag-handle {
font-size: 1.4em; margin-left: auto;
color: #6dd6f4; font-size: 1.4em;
opacity: 0.4; color: #6dd6f4;
user-select: none; opacity: 0.4;
transition: opacity 0.2s ease; user-select: none;
flex-shrink: 0; transition: opacity 0.2s ease;
flex-shrink: 0;
} }
.cluster-draggable:hover .cluster-drag-handle { .cluster-draggable:hover .cluster-drag-handle {
opacity: 0.8; opacity: 0.8;
} }
.cohesion-score { .cohesion-score {
text-align: center; text-align: center;
font-size: 0.95em; font-size: 0.95em;
color: #e0e0e0; color: #e0e0e0;
padding: 8px 16px; padding: 8px 16px;
background: rgba(109, 214, 244, 0.1); background: rgba(109, 214, 244, 0.1);
border-radius: 6px; border-radius: 6px;
width: fit-content; width: fit-content;
margin: 0 auto; margin: 0 auto;
} }

View File

@@ -14,7 +14,7 @@ const MOCK_CLUSTER_RESPONSE = {
{ label: 'Auth Issues', noteIds: ['note_001'] }, { label: 'Auth Issues', noteIds: ['note_001'] },
{ label: 'Export Issues', noteIds: ['note_002'] }, { label: 'Export Issues', noteIds: ['note_002'] },
], ],
score: 0.74, score: 0.09,
}; };
let fetchMock: ReturnType<typeof vi.fn>; let fetchMock: ReturnType<typeof vi.fn>;