adding bibtex citation of articles

src/components/DatasetMetadataModal.module.css

@@ -324,6 +324,18 @@
 	background: var(--agml-surface-soft);
 }
 
+.snippetLabel {
+	font-family: var(--ifm-code-font-family);
+	font-size: 0.68rem;
+	font-weight: 600;
+	color: var(--agml-muted);
+	margin: 0.9rem 0 0;
+}
+
+.snippetLabel:first-of-type {
+	margin-top: 0;
+}
+
 .snippetRow {
 	display: flex;
 	align-items: flex-start;

src/components/DatasetMetadataModal.tsx

@@ -113,6 +113,12 @@ function formatLoaderInstructions(dataset: Dataset) {
 				code: `from datasets import load_dataset\nloader = load_dataset("Project-AgML/${dataset.name}")`,
 			};
 		}
+		if (dataset.sensor_modality?.toLowerCase().includes('multispectral')) {
+			return {
+				title: 'Load from Hugging Face',
+				code: `from agml.data.multispectral_hf_loader import MultispectralDataLoader\nloader = MultispectralDataLoader("Project-AgML/${dataset.name}")`,
+			};
+		}
 		return {
 			title: 'Load from Hugging Face',
 			code: `from agml.data.hf_loader import HuggingFaceDataLoader\nloader = HuggingFaceDataLoader("Project-AgML/${dataset.name}")`,
@@ -1166,6 +1172,7 @@ export function DatasetMetadataModal({
 	const [classesExpanded, setClassesExpanded] = useState(false);
 	const [copied, setCopied] = useState(false);
 	const [citationCopied, setCitationCopied] = useState(false);
+	const [bibtexCopied, setBibtexCopied] = useState(false);
 
 	const [expandedRowKeys, setExpandedRowKeys] = useState<Set<string>>(new Set());
 	const toggleExpandedRow = (key: string) => {
@@ -1595,23 +1602,47 @@ export function DatasetMetadataModal({
 					)}
 				</section>
 
-				{dataset.citation && (
+				{(dataset.citation || dataset.bibtex) && (
 					<section className={styles.section}>
 						<h3 className={styles.sectionTitle}>Citation</h3>
-						<div className={styles.snippetRow}>
-							<pre className={styles.citationCode}>{dataset.citation}</pre>
-							<button
-								type="button"
-								className={styles.snippetCopyButton}
-								onClick={() => {
-									navigator.clipboard.writeText(dataset.citation ?? '');
-									setCitationCopied(true);
-									setTimeout(() => setCitationCopied(false), 1500);
-								}}
-							>
-								{citationCopied ? 'Copied!' : 'Copy'}
-							</button>
-						</div>
+						{dataset.citation && (
+							<>
+								{dataset.bibtex && <p className={styles.snippetLabel}>Citation</p>}
+								<div className={styles.snippetRow}>
+									<pre className={styles.citationCode}>{dataset.citation}</pre>
+									<button
+										type="button"
+										className={styles.snippetCopyButton}
+										onClick={() => {
+											navigator.clipboard.writeText(dataset.citation ?? '');
+											setCitationCopied(true);
+											setTimeout(() => setCitationCopied(false), 1500);
+										}}
+									>
+										{citationCopied ? 'Copied!' : 'Copy'}
+									</button>
+								</div>
+							</>
+						)}
+						{dataset.bibtex && (
+							<>
+								{dataset.citation && <p className={styles.snippetLabel}>BibTeX</p>}
+								<div className={styles.snippetRow}>
+									<pre className={styles.citationCode}>{dataset.bibtex}</pre>
+									<button
+										type="button"
+										className={styles.snippetCopyButton}
+										onClick={() => {
+											navigator.clipboard.writeText(dataset.bibtex ?? '');
+											setBibtexCopied(true);
+											setTimeout(() => setBibtexCopied(false), 1500);
+										}}
+									>
+										{bibtexCopied ? 'Copied!' : 'Copy'}
+									</button>
+								</div>
+							</>
+						)}
 					</section>
 				)}
 				</div>

src/lib/datasets.ts

@@ -19,6 +19,10 @@ export interface Dataset {
     augmented_counterpart: string | null;
     crop_types: string[] | null;
     sensor_modality: string | null;
+    // Derived: `sensor_modality` split into individual lowercase values (it's stored as a single
+    // comma-joined string, e.g. "rgb, thermal") so the search page can filter by modality even
+    // when a dataset has more than one.
+    display_modality: string[] | null;
     real_or_synthetic: string | null;
     platform: string[] | null;
     input_data_format: string | null;
@@ -34,6 +38,10 @@ export interface Dataset {
     point_cloud_sample_url: string | null;
     license: string | null;
     citation: string | null;
+    // BibTeX entry for the article the dataset was sourced from, scraped from the dataset's
+    // Hugging Face README (or resolved from a DOI found there) — distinct from `citation`,
+    // which is free-form and not always valid BibTeX.
+    bibtex: string | null;
     parent_dataset?: string | null;
     // Derived (not present in the source manifests): for a dataset whose child variants are
     // hidden from search results (see isChildDataset/includeChildren), the crop/species names
@@ -158,6 +166,10 @@ function normalizeDataset(raw: unknown): Dataset | null {
         raw.ml_task,
         raw.task,
     );
+    const sensorModality =
+        firstString(raw.sensor_modality, raw.sensor, raw.modality) ??
+        toStringArray(raw.sensor_modality)?.join(", ") ??
+        null;
     const qaType = toStringArray(raw.qa_type);
     const taskDimensions = toStringArray(raw.task_dimensions);
     const isVlm =
@@ -188,10 +200,17 @@ function normalizeDataset(raw: unknown): Dataset | null {
                 ?.map((c) => c.trim().toLowerCase())
                 .filter((c) => c && !/^\+.*\b(more|others?)\b/i.test(c)) ??
             null,
-        sensor_modality:
-            firstString(raw.sensor_modality, raw.sensor, raw.modality) ??
-            toStringArray(raw.sensor_modality)?.join(", ") ??
-            null,
+        sensor_modality: sensorModality,
+        display_modality: sensorModality
+            ? Array.from(
+                  new Set(
+                      sensorModality
+                          .split(",")
+                          .map((entry) => entry.trim().toLowerCase())
+                          .filter(Boolean),
+                  ),
+              )
+            : null,
         real_or_synthetic: firstString(
             raw.real_or_synthetic,
             raw.real_synthetic,
@@ -225,6 +244,7 @@ function normalizeDataset(raw: unknown): Dataset | null {
         ),
         license: firstString(raw.license),
         citation: firstString(raw.citation),
+        bibtex: firstString(raw.bibtex),
         parent_dataset: firstString(raw.parent_dataset, raw.parentDataset),
         child_crop_types: null,
         zip_size_bytes: toNumber(raw.zip_size_bytes ?? raw.zipSizeBytes),
@@ -261,6 +281,7 @@ function mergeDataset(current: Dataset, incoming: Dataset): Dataset {
             current.augmented_counterpart ?? incoming.augmented_counterpart,
         crop_types: current.crop_types ?? incoming.crop_types,
         sensor_modality: current.sensor_modality ?? incoming.sensor_modality,
+        display_modality: current.display_modality ?? incoming.display_modality,
         real_or_synthetic:
             current.real_or_synthetic ?? incoming.real_or_synthetic,
         platform:
@@ -292,6 +313,7 @@ function mergeDataset(current: Dataset, incoming: Dataset): Dataset {
             current.point_cloud_sample_url ?? incoming.point_cloud_sample_url,
         license: current.license ?? incoming.license,
         citation: current.citation ?? incoming.citation,
+        bibtex: current.bibtex ?? incoming.bibtex,
         parent_dataset: current.parent_dataset ?? incoming.parent_dataset,
         child_crop_types: current.child_crop_types ?? incoming.child_crop_types,
         zip_size_bytes: current.zip_size_bytes ?? incoming.zip_size_bytes,

src/pages/datasets/index.tsx

@@ -66,6 +66,14 @@ const DATASET_FILTERS: DatasetFilterConfig[] = [
     mode: 'containsAny',
     formatOption: (value) => toDisplayLabel(value),
   },
+  {
+    key: 'modality',
+    label: 'Modality',
+    field: 'display_modality',
+    kind: 'dropdown',
+    mode: 'containsAny',
+    formatOption: (value) => toDisplayLabel(value),
+  },
   {
     key: 'environment',
     label: 'Environment',