diff --git a/ui/src/frontend/base_slice_track.ts b/ui/src/frontend/base_slice_track.ts
index 0ca6c01..45cbc9e 100644
--- a/ui/src/frontend/base_slice_track.ts
+++ b/ui/src/frontend/base_slice_track.ts
@@ -35,6 +35,7 @@
 import {TrackMouseEvent, TrackRenderContext} from '../public/track';
 import {Point2D, VerticalBounds} from '../base/geom';
 import {Trace} from '../public/trace';
+import {Ds} from '../trace_processor/dataset';
 
 // The common class that underpins all tracks drawing slices.
 
@@ -972,6 +973,17 @@
     });
     return {ts: Time.fromRaw(row.ts), dur: Duration.fromRaw(row.dur)};
   }
+
+  getDataset(): Ds.Dataset | undefined {
+    return {
+      src: this.getSqlSource(),
+      schema: {
+        id: NUM,
+        ts: LONG,
+        dur: LONG,
+      },
+    };
+  }
 }
 
 // This is the argument passed to onSliceOver(args).
diff --git a/ui/src/frontend/named_slice_track.ts b/ui/src/frontend/named_slice_track.ts
index ed9b5f0..edb7ec9 100644
--- a/ui/src/frontend/named_slice_track.ts
+++ b/ui/src/frontend/named_slice_track.ts
@@ -16,7 +16,7 @@
 import {TrackEventDetailsPanel} from '../public/details_panel';
 import {TrackEventSelection} from '../public/selection';
 import {Slice} from '../public/track';
-import {STR_NULL} from '../trace_processor/query_result';
+import {LONG, NUM, STR, STR_NULL} from '../trace_processor/query_result';
 import {
   BASE_ROW,
   BaseSliceTrack,
@@ -30,6 +30,7 @@
 import {renderDuration} from './widgets/duration';
 import {TraceImpl} from '../core/trace_impl';
 import {assertIsInstance} from '../base/logging';
+import {Ds} from '../trace_processor/dataset';
 
 export const NAMED_ROW = {
   // Base columns (tsq, ts, dur, id, depth).
@@ -80,4 +81,16 @@
     // because this class is exposed to plugins (which see only Trace).
     return new ThreadSliceDetailsPanel(assertIsInstance(this.trace, TraceImpl));
   }
+
+  override getDataset(): Ds.Dataset | undefined {
+    return {
+      src: this.getSqlSource(),
+      schema: {
+        id: NUM,
+        name: STR,
+        ts: LONG,
+        dur: LONG,
+      },
+    };
+  }
 }
diff --git a/ui/src/plugins/dev.perfetto.AsyncSlices/async_slice_track.ts b/ui/src/plugins/dev.perfetto.AsyncSlices/async_slice_track.ts
index 1bb31a5..4c8898b 100644
--- a/ui/src/plugins/dev.perfetto.AsyncSlices/async_slice_track.ts
+++ b/ui/src/plugins/dev.perfetto.AsyncSlices/async_slice_track.ts
@@ -14,12 +14,19 @@
 
 import {BigintMath as BIMath} from '../../base/bigint_math';
 import {clamp} from '../../base/math_utils';
+import {Ds} from '../../trace_processor/dataset';
 import {NAMED_ROW, NamedSliceTrack} from '../../frontend/named_slice_track';
 import {SLICE_LAYOUT_FIT_CONTENT_DEFAULTS} from '../../frontend/slice_layout';
 import {NewTrackArgs} from '../../frontend/track';
 import {TrackEventDetails} from '../../public/selection';
 import {Slice} from '../../public/track';
-import {LONG_NULL} from '../../trace_processor/query_result';
+import {
+  LONG,
+  LONG_NULL,
+  NUM,
+  NUM_NULL,
+  STR,
+} from '../../trace_processor/query_result';
 
 export const THREAD_SLICE_ROW = {
   // Base columns (tsq, ts, dur, id, depth).
@@ -104,4 +111,21 @@
       tableName: 'slice',
     };
   }
+
+  override getDataset(): Ds.Dataset {
+    return {
+      src: `slice`,
+      filter: {
+        col: 'track_id',
+        in: this.trackIds,
+      },
+      schema: {
+        id: NUM,
+        name: STR,
+        ts: LONG,
+        dur: LONG,
+        parent_id: NUM_NULL,
+      },
+    };
+  }
 }
diff --git a/ui/src/plugins/dev.perfetto.AsyncSlices/slice_selection_aggregator.ts b/ui/src/plugins/dev.perfetto.AsyncSlices/slice_selection_aggregator.ts
index 55f7c95..0364152 100644
--- a/ui/src/plugins/dev.perfetto.AsyncSlices/slice_selection_aggregator.ts
+++ b/ui/src/plugins/dev.perfetto.AsyncSlices/slice_selection_aggregator.ts
@@ -16,16 +16,27 @@
 import {AreaSelection} from '../../public/selection';
 import {Engine} from '../../trace_processor/engine';
 import {AreaSelectionAggregator} from '../../public/selection';
-import {SLICE_TRACK_KIND} from '../../public/track_kinds';
+import {Ds} from '../../trace_processor/dataset';
+import {LONG, NUM, STR} from '../../trace_processor/query_result';
 
 export class SliceSelectionAggregator implements AreaSelectionAggregator {
   readonly id = 'slice_aggregation';
 
   async createAggregateView(engine: Engine, area: AreaSelection) {
-    const selectedTrackKeys = getSelectedTrackSqlIds(area);
-
-    if (selectedTrackKeys.length === 0) return false;
-
+    const desiredSchema = {
+      id: NUM,
+      name: STR,
+      ts: LONG,
+      dur: LONG,
+    };
+    const validDatasets = area.tracks
+      .map((t) => t.track.getDataset?.())
+      .filter((d) => d !== undefined)
+      .filter((d) => Ds.doesImplement(d, desiredSchema));
+    if (validDatasets.length === 0) {
+      return false;
+    }
+    const optimizedDataset = Ds.optimize({union: validDatasets});
     await engine.query(`
       create or replace perfetto table ${this.id} as
       select
@@ -33,12 +44,13 @@
         sum(dur) AS total_dur,
         sum(dur)/count() as avg_dur,
         count() as occurrences
-        from slices
-      where track_id in (${selectedTrackKeys})
-        and ts + dur > ${area.start}
+        from (${Ds.query(optimizedDataset)})
+      where
+        ts + dur > ${area.start}
         and ts < ${area.end}
       group by name
     `);
+
     return true;
   }
 
@@ -83,14 +95,3 @@
     ];
   }
 }
-
-function getSelectedTrackSqlIds(area: AreaSelection): number[] {
-  const selectedTrackKeys: number[] = [];
-  for (const trackInfo of area.tracks) {
-    if (trackInfo?.tags?.kind === SLICE_TRACK_KIND) {
-      trackInfo.tags.trackIds &&
-        selectedTrackKeys.push(...trackInfo.tags.trackIds);
-    }
-  }
-  return selectedTrackKeys;
-}
diff --git a/ui/src/plugins/dev.perfetto.Frames/actual_frames_track.ts b/ui/src/plugins/dev.perfetto.Frames/actual_frames_track.ts
index d75dd77..2c19110 100644
--- a/ui/src/plugins/dev.perfetto.Frames/actual_frames_track.ts
+++ b/ui/src/plugins/dev.perfetto.Frames/actual_frames_track.ts
@@ -102,6 +102,15 @@
       tableName: 'slice',
     };
   }
+
+  // Override dataset from base class NamedSliceTrack as we don't want these
+  // tracks to participate in generic area selection aggregation (frames tracks
+  // have their own dedicated aggregation panel).
+  // TODO(stevegolton): In future CLs this will be handled with aggregation keys
+  // instead, as this track will have to expose a dataset anyway.
+  override getDataset() {
+    return undefined;
+  }
 }
 
 function getColorSchemeForJank(
diff --git a/ui/src/plugins/dev.perfetto.Ftrace/ftrace_track.ts b/ui/src/plugins/dev.perfetto.Ftrace/ftrace_track.ts
index 78c59c8..d056a8d 100644
--- a/ui/src/plugins/dev.perfetto.Ftrace/ftrace_track.ts
+++ b/ui/src/plugins/dev.perfetto.Ftrace/ftrace_track.ts
@@ -20,10 +20,11 @@
 import {TrackData} from '../../common/track_data';
 import {Engine} from '../../trace_processor/engine';
 import {Track} from '../../public/track';
-import {LONG, STR} from '../../trace_processor/query_result';
+import {LONG, NUM, STR} from '../../trace_processor/query_result';
 import {FtraceFilter} from './common';
 import {Monitor} from '../../base/monitor';
 import {TrackRenderContext} from '../../public/track';
+import {Ds} from '../../trace_processor/dataset';
 
 const MARGIN = 2;
 const RECT_HEIGHT = 18;
@@ -56,6 +57,25 @@
     this.monitor = new Monitor([() => store.state]);
   }
 
+  getDataset(): Ds.Dataset {
+    return {
+      // 'ftrace_event' doesn't have a dur column, but injecting dur=0 (all
+      // ftrace events are effectively 'instant') allows us to participate in
+      // generic slice aggregations
+      src: 'select id, ts, 0 as dur, name from ftrace_event',
+      schema: {
+        id: NUM,
+        name: STR,
+        ts: LONG,
+        dur: LONG,
+      },
+      filter: {
+        col: 'cpu',
+        eq: this.cpu,
+      },
+    };
+  }
+
   async onUpdate({
     visibleWindow,
     resolution,
diff --git a/ui/src/public/track.ts b/ui/src/public/track.ts
index 94ac9e7..93d493e 100644
--- a/ui/src/public/track.ts
+++ b/ui/src/public/track.ts
@@ -20,6 +20,7 @@
 import {ColorScheme} from './color_scheme';
 import {TrackEventDetailsPanel} from './details_panel';
 import {TrackEventDetails, TrackEventSelection} from './selection';
+import {Ds} from '../trace_processor/dataset';
 
 export interface TrackManager {
   /**
@@ -175,6 +176,12 @@
   onMouseOut?(): void;
 
   /**
+   * Optional: Returns a dataset that represents the events displayed on this
+   * track.
+   */
+  getDataset?(): Ds.Dataset | undefined;
+
+  /**
    * Optional: Get details of a track event given by eventId on this track.
    */
   getSelectionDetails?(eventId: number): Promise<TrackEventDetails | undefined>;
diff --git a/ui/src/trace_processor/dataset.ts b/ui/src/trace_processor/dataset.ts
new file mode 100644
index 0000000..6863570
--- /dev/null
+++ b/ui/src/trace_processor/dataset.ts
@@ -0,0 +1,258 @@
+// Copyright (C) 2024 The Android Open Source Project
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+import {assertUnreachable} from '../base/logging';
+import {getOrCreate} from '../base/utils';
+import {ColumnType, SqlValue} from './query_result';
+
+export namespace Ds {
+  export type Dataset = UnionDataset | SourceDataset;
+  export type Schema = Record<string, ColumnType>;
+
+  /**
+   * Defines a dataset with a source SQL select statement of table name, a
+   * schema describing the columns, and an optional filter.
+   */
+  export interface SourceDataset {
+    readonly src: string;
+    readonly schema: Schema;
+    readonly filter?: EqFilter | InFilter;
+  }
+
+  /**
+   * A dataset that represents the union of multiple datasets.
+   */
+  export interface UnionDataset {
+    readonly union: ReadonlyArray<Dataset>;
+  }
+
+  /**
+   * Generic filter type.
+   */
+  export type Filter = EqFilter | InFilter;
+
+  /**
+   * A filter used to express that a column must equal a value.
+   */
+  export interface EqFilter {
+    readonly col: string;
+    readonly eq: SqlValue;
+  }
+
+  /**
+   * A filter used to express that column must be one of a set of values.
+   */
+  export interface InFilter {
+    readonly col: string;
+    readonly in: ReadonlyArray<SqlValue>;
+  }
+
+  /**
+   * Returns true if the dataset implements a given schema.
+   *
+   * Note: `implements` is a reserved keyword in TS so we can't call this
+   * function `implements`.
+   *
+   * @param dataset - The dataset to test.
+   * @param testSchema - The schema to test against.
+   */
+  export function doesImplement(dataset: Dataset, testSchema: Schema): boolean {
+    const datasetSchema = schema(dataset);
+    return Object.entries(testSchema).every(([name, kind]) => {
+      return name in datasetSchema && datasetSchema[name] === kind;
+    });
+  }
+
+  /**
+   * This function optimizes a dataset into the smallest possible expression.
+   *
+   * For example by combining elements of union data sets that have the same src
+   * and similar filters into a single set.
+   *
+   * For example, the following union data set...
+   *
+   * ```
+   * {
+   *   union: [
+   *     {
+   *       src: 'foo',
+   *       schema: {
+   *         'a': NUM,
+   *         'b': NUM,
+   *       },
+   *       filter: {col: 'a', eq: 1},
+   *     },
+   *     {
+   *       src: 'foo',
+   *       schema: {
+   *         'a': NUM,
+   *         'b': NUM,
+   *       },
+   *       filter: {col: 'a', eq: 2},
+   *     },
+   *   ]
+   * }
+   * ```
+   *
+   * ...will be combined into a single set...
+   *
+   * ```
+   * {
+   *   src: 'foo',
+   *   schema: {
+   *     'a': NUM,
+   *     'b': NUM,
+   *   },
+   *   filter: {col: 'a', in: [1, 2]},
+   * },
+   * ```
+   *
+   * @param dataset - The dataset to optimize.
+   */
+  export function optimize(dataset: Dataset): Dataset {
+    if ('src' in dataset) {
+      // No optimization possible for individual datasets
+      return dataset;
+    } else if ('union' in dataset) {
+      // Recursively optimize each dataset of this union
+      const optimizedUnion = dataset.union.map(optimize);
+
+      // Find all source datasets and combine then based on src
+      const combinedSrcSets = new Map<string, SourceDataset[]>();
+      const otherDatasets: Dataset[] = [];
+      for (const e of optimizedUnion) {
+        if ('src' in e) {
+          const set = getOrCreate(combinedSrcSets, e.src, () => []);
+          set.push(e);
+        } else {
+          otherDatasets.push(e);
+        }
+      }
+
+      const mergedSrcSets = Array.from(combinedSrcSets.values()).map(
+        (srcGroup) => {
+          if (srcGroup.length === 1) return srcGroup[0];
+
+          // Combine schema across all members in the union
+          const combinedSchema = srcGroup.reduce((acc, e) => {
+            Object.assign(acc, e.schema);
+            return acc;
+          }, {} as Schema);
+
+          // Merge filters for the same src
+          const inFilters: InFilter[] = [];
+          for (const {filter} of srcGroup) {
+            if (filter) {
+              if ('eq' in filter) {
+                inFilters.push({col: filter.col, in: [filter.eq]});
+              } else {
+                inFilters.push(filter);
+              }
+            }
+          }
+
+          const mergedFilter = mergeFilters(inFilters);
+          return {
+            src: srcGroup[0].src,
+            schema: combinedSchema,
+            filter: mergedFilter,
+          };
+        },
+      );
+
+      const finalUnion = [...mergedSrcSets, ...otherDatasets];
+
+      if (finalUnion.length === 1) {
+        return finalUnion[0];
+      } else {
+        return {union: finalUnion};
+      }
+    } else {
+      assertUnreachable(dataset);
+    }
+  }
+
+  function mergeFilters(filters: InFilter[]): InFilter | undefined {
+    if (filters.length === 0) return undefined;
+    const col = filters[0].col;
+    const values = new Set(filters.flatMap((filter) => filter.in));
+    return {col, in: Array.from(values)};
+  }
+
+  /**
+   * Get the schema of an dataset.
+   *
+   * @param dataset - The dataset to get the schema of.
+   */
+  export function schema(dataset: Dataset): Schema {
+    if ('src' in dataset) {
+      return dataset.schema;
+    } else if ('union' in dataset) {
+      // Find the minimal set of columns that are supported by all datasets of
+      // the union
+      let sch: Record<string, ColumnType> | undefined = undefined;
+      dataset.union.forEach((e) => {
+        const eSchema = schema(e);
+        if (sch === undefined) {
+          // First time just use this one
+          sch = eSchema;
+        } else {
+          const newSch: Record<string, ColumnType> = {};
+          for (const [key, kind] of Object.entries(sch)) {
+            if (key in eSchema && eSchema[key] === kind) {
+              newSch[key] = kind;
+            }
+          }
+          sch = newSch;
+        }
+      });
+      return sch ?? {};
+    } else {
+      assertUnreachable(dataset);
+    }
+  }
+
+  /**
+   * Produce a query for this dataset.
+   *
+   * @param dataset - The dataset to get the query for.
+   * @param sch - The schema to use for extracting columns - if undefined, the
+   * most specific possible schema is evaluated from the dataset first and used
+   * instead.
+   */
+  export function query(dataset: Dataset, sch?: Schema): string {
+    function filterToQuery(filter: Filter) {
+      if ('eq' in filter) {
+        return `where ${filter.col} = ${filter.eq}`;
+      } else if ('in' in filter) {
+        return `where ${filter.col} in (${filter.in.join(',')})`;
+      } else {
+        assertUnreachable(filter);
+      }
+    }
+
+    sch = sch ?? schema(dataset);
+    if ('src' in dataset) {
+      const whereClause = dataset.filter ? filterToQuery(dataset.filter) : '';
+      const cols = Object.keys(sch);
+      return `select ${cols.join(', ')} from (${dataset.src}) ${whereClause}`.trim();
+    } else if ('union' in dataset) {
+      return dataset.union
+        .map((dataset) => query(dataset, sch))
+        .join(' union all ');
+    } else {
+      assertUnreachable(dataset);
+    }
+  }
+}
diff --git a/ui/src/trace_processor/dataset_unittest.ts b/ui/src/trace_processor/dataset_unittest.ts
new file mode 100644
index 0000000..e354b54
--- /dev/null
+++ b/ui/src/trace_processor/dataset_unittest.ts
@@ -0,0 +1,242 @@
+// Copyright (C) 2024 The Android Open Source Project
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//      http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+import {Ds} from './dataset';
+import {LONG, NUM, STR} from './query_result';
+
+test('get query for simple dataset', () => {
+  const dataset: Ds.Dataset = {
+    src: 'slice',
+    schema: {id: NUM},
+  };
+
+  expect(Ds.query(dataset)).toEqual('select id from (slice)');
+});
+
+test("get query for simple dataset with 'eq' filter", () => {
+  const dataset: Ds.Dataset = {
+    src: 'slice',
+    schema: {id: NUM},
+    filter: {
+      col: 'id',
+      eq: 123,
+    },
+  };
+
+  expect(Ds.query(dataset)).toEqual('select id from (slice) where id = 123');
+});
+
+test("get query for simple dataset with an 'in' filter", () => {
+  const dataset: Ds.Dataset = {
+    src: 'slice',
+    schema: {id: NUM},
+    filter: {
+      col: 'id',
+      in: [123, 456],
+    },
+  };
+
+  expect(Ds.query(dataset)).toEqual(
+    'select id from (slice) where id in (123,456)',
+  );
+});
+
+test('get query for union dataset', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {id: NUM},
+        filter: {
+          col: 'id',
+          eq: 123,
+        },
+      },
+      {
+        src: 'slice',
+        schema: {id: NUM},
+        filter: {
+          col: 'id',
+          eq: 456,
+        },
+      },
+    ],
+  };
+
+  expect(Ds.query(dataset)).toEqual(
+    'select id from (slice) where id = 123 union all select id from (slice) where id = 456',
+  );
+});
+
+test('doesImplement', () => {
+  const dataset = {
+    src: 'slice',
+    schema: {id: NUM, ts: LONG},
+  };
+
+  expect(Ds.doesImplement(dataset, {id: NUM})).toBe(true);
+  expect(Ds.doesImplement(dataset, {id: NUM, ts: LONG})).toBe(true);
+  expect(Ds.doesImplement(dataset, {id: NUM, ts: LONG, name: STR})).toBe(false);
+  expect(Ds.doesImplement(dataset, {id: LONG})).toBe(false);
+});
+
+test('find the schema of a simple dataset', () => {
+  const dataset: Ds.Dataset = {
+    src: 'slice',
+    schema: {id: NUM, ts: LONG},
+  };
+
+  expect(Ds.schema(dataset)).toMatchObject({id: NUM, ts: LONG});
+});
+
+test('find the schema of a union where source sets differ in their names', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {foo: NUM},
+      },
+      {
+        src: 'slice',
+        schema: {bar: NUM},
+      },
+    ],
+  };
+
+  expect(Ds.schema(dataset)).toMatchObject({});
+});
+
+test('find the schema of a union with differing source sets', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {foo: NUM},
+      },
+      {
+        src: 'slice',
+        schema: {foo: LONG},
+      },
+    ],
+  };
+
+  expect(Ds.schema(dataset)).toMatchObject({});
+});
+
+test('find the schema of a union with one column in common', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {foo: NUM, bar: NUM},
+      },
+      {
+        src: 'slice',
+        schema: {foo: NUM, baz: NUM},
+      },
+    ],
+  };
+
+  expect(Ds.schema(dataset)).toMatchObject({foo: NUM});
+});
+
+test('optimize a union dataset', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {},
+        filter: {
+          col: 'track_id',
+          eq: 123,
+        },
+      },
+      {
+        src: 'slice',
+        schema: {},
+        filter: {
+          col: 'track_id',
+          eq: 456,
+        },
+      },
+    ],
+  };
+
+  expect(Ds.optimize(dataset)).toEqual({
+    src: 'slice',
+    schema: {},
+    filter: {
+      col: 'track_id',
+      in: [123, 456],
+    },
+  });
+});
+
+test('optimize a union dataset with different types of filters', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {},
+        filter: {
+          col: 'track_id',
+          eq: 123,
+        },
+      },
+      {
+        src: 'slice',
+        schema: {},
+        filter: {
+          col: 'track_id',
+          in: [456, 789],
+        },
+      },
+    ],
+  };
+
+  expect(Ds.optimize(dataset)).toEqual({
+    src: 'slice',
+    schema: {},
+    filter: {
+      col: 'track_id',
+      in: [123, 456, 789],
+    },
+  });
+});
+
+test('optimize a union dataset with different schemas', () => {
+  const dataset: Ds.Dataset = {
+    union: [
+      {
+        src: 'slice',
+        schema: {foo: NUM},
+      },
+      {
+        src: 'slice',
+        schema: {bar: NUM},
+      },
+    ],
+  };
+
+  expect(Ds.optimize(dataset)).toEqual({
+    src: 'slice',
+    // The resultant schema is the combination of the union's member's schemas,
+    // as we know the source is the same as we know we can get all of the 'seen'
+    // columns from the source.
+    schema: {
+      foo: NUM,
+      bar: NUM,
+    },
+  });
+});