Added basic entities view

Author: PrinsINT

client/components.d.ts

@@ -18,6 +18,7 @@ declare module 'vue' {
     CorpusTable: typeof import('./src/components/tables/CorpusTable.vue')['default']
     DeleteModal: typeof import('./src/components/modals/DeleteModal.vue')['default']
     DifferentTagsetsHelp: typeof import('./src/components/help/DifferentTagsetsHelp.vue')['default']
+    DocumentSelect: typeof import('./src/components/input/DocumentSelect.vue')['default']
     DocumentsHelp: typeof import('./src/components/help/DocumentsHelp.vue')['default']
     DocumentsTable: typeof import('./src/components/tables/DocumentsTable.vue')['default']
     DownloadButton: typeof import('./src/components/input/DownloadButton.vue')['default']

client/src/api/evaluation.ts

@@ -8,13 +8,15 @@ import type { UUID } from "@/types/corpora"
 import type {
     ConfusionWrapper,
     DistributionWrapper,
+    Entity,
     Metrics,
     TermComparison,
 } from "@/types/evaluation"
 
 type ConfusionResponse = AxiosResponse<ConfusionWrapper>
 type DistributionResponse = AxiosResponse<DistributionWrapper>
 type MetricsResponse = AxiosResponse<Metrics>
+type EntitiesResponse = AxiosResponse<Entity[]>
 
 const evaluationPath = (corpus: UUID, hypothesis: string): string =>
     `/corpora/${corpus}/jobs/${hypothesis}/evaluation`
@@ -35,6 +37,8 @@ const documentLayerComparisonPath = (
     job: string,
     document: string,
 ): string => `/corpora/${corpus}/jobs/${job}/documents/${document}/evaluation`
+const entitiesPath = (corpus: UUID, job: string, document: string): string =>
+    `/corpora/${corpus}/jobs/${job}/documents/${document}/entities`
 
 /**
  * Fetch term frequency distribution.
@@ -161,3 +165,11 @@ export function getDocumentLayerComparison(
         params: { reference },
     })
 }
+
+export function getEntities(
+    corpus: UUID,
+    job: string,
+    document: string,
+): Promise<EntitiesResponse> {
+    return axios.get(entitiesPath(corpus, job, document))
+}

client/src/components/input/DocumentSelect.vue

@@ -0,0 +1,20 @@
+<template>
+    <label for="document-select">Document:</label>
+    <GSelect v-model="model" :options id="document-select"/>
+</template>
+
+<script setup lang="ts">
+import stores from "@/stores"
+import type { SelectOption } from "@/types/ui/select"
+
+const documentsStore = stores.useDocuments()
+
+const model = defineModel<string>()
+
+const options = computed<SelectOption[]>(() => {
+    return documentsStore.available.map(doc => ({
+        value: doc.name,
+        text: doc.name,
+    }))
+})
+</script>

client/src/components/input/JobSelect.vue

@@ -49,21 +49,25 @@ const selectedJob = computed<string>({
 
 // Whether there are documents that have not been tagged yet.
 // Not relevant for source layer.
-const untaggedDocsExist = computed(() => {
-    if (!selectedJob.value) return false
-    if (!jobsStore.jobs) return false
-    const job = jobsStore.jobs[selectedJob.value]
-    if (!job) return false
-    if (job.tagger.id === SOURCE_LAYER) return false
-    return job.progress.finished < job.progress.total
+const untaggedDocsExist = computed<boolean>(() => {
+    if (!job.value) return false
+    return job.value.progress.finished < job.value.progress.total
 })
 // Whether the selected layer is sourceLayer and has missing annotations.
-const sourceLayerHasMissingAnnotations = computed(() => {
-    if (!selectedJob.value) return false
-    if (!jobsStore.jobs) return false
-    const job = jobsStore.jobs[selectedJob.value]
-    if (!job) return false
-    if (job.tagger.id !== SOURCE_LAYER) return false
-    return job.progress.finished !== documentsStore.numSourceAnnotations
+const sourceLayerHasMissingAnnotations = computed<boolean>(() => {
+    if (!job.value) return false
+    return job.value.progress.finished !== documentsStore.numSourceAnnotations
+})
+
+const job = computed<Job | undefined>(() => {
+    if (!selectedJob.value) return undefined
+    if (!jobsStore.jobs) return undefined
+    // any job could be any job, even the source layer
+    const anyJob = jobsStore.jobs[selectedJob.value]
+    if (!anyJob) return undefined
+    // if the job is not source layer, return it
+    if (anyJob.tagger.id !== SOURCE_LAYER) return anyJob
+    // if the job is source layer, return undefined
+    return undefined
 })
 </script>

client/src/router/index.ts

@@ -9,6 +9,7 @@ import EvaluateView from "@/views/annotate/subviews/evaluate/EvaluateView.vue"
 import ConfusionView from "@/views/annotate/subviews/evaluate/subviews/ConfusionView.vue"
 import DistributionView from "@/views/annotate/subviews/evaluate/subviews/DistributionView.vue"
 import DocumentLayerComparisonView from "@/views/annotate/subviews/evaluate/subviews/DocumentLayerComparisonView.vue"
+import EntitiesView from "@/views/annotate/subviews/evaluate/subviews/EntitiesView.vue"
 import GlobalMetricsView from "@/views/annotate/subviews/evaluate/subviews/GlobalMetricsView.vue"
 import GroupedMetricsView from "@/views/annotate/subviews/evaluate/subviews/GroupedMetricsView.vue"
 import ExportView from "@/views/annotate/subviews/ExportView.vue"
@@ -94,6 +95,11 @@ const routes = [
                         path: "document_layer_comparison",
                         component: DocumentLayerComparisonView,
                     },
+                    {
+                        meta: { title: "Evaluate" },
+                        path: "entities",
+                        component: EntitiesView,
+                    },
                 ],
             },
             {

client/src/types/evaluation.ts

@@ -71,8 +71,11 @@ export type TermComparison = {
     partialOverlap: boolean
 }
 
-export interface Term {
+export type Term = {
+    id: string
+    offset: number
     annotations: Record<string, string>
+    spaceAfter?: boolean
 }
 
 export type WordForm = {
@@ -86,3 +89,8 @@ export type EvaluationEntry = {
     count: number
     samples: TermComparison[]
 }
+
+export type Entity = {
+    first: string
+    second: Term[]
+}

client/src/views/annotate/subviews/evaluate/EvaluateView.vue

@@ -27,6 +27,7 @@
             { id: 'grouped_metrics', title: 'Grouped Metrics' },
             { id: 'confusion', title: 'Pos Confusion' },
             { id: 'document_layer_comparison', title: 'Document View' },
+            { id: 'entities', title: 'Entities View' },
         ]">
         </GTabs>
     </AnnotateTab>

client/src/views/annotate/subviews/evaluate/subviews/EntitiesView.vue

@@ -0,0 +1,63 @@
+<template>
+    <GCard title="Entities View">
+        <template #help>
+            <p>
+                Here you can see all the named entities in the selected document.
+            </p>
+        </template>
+        <DocumentSelect v-if="jobSelection.hypothesisJobId" v-model="selectedDoc" />
+        <p v-else>
+            Please select a hypothesis job first.
+        </p>
+        <GTable compact v-if="items" :title :loading :items :columns>
+            <template #table-empty-instruction>
+                No entities found in this document.
+            </template>
+
+            <template #cell-second="data: TableData<Term>">
+                {{ displayEntity(data.value) }}
+            </template>
+        </GTable>
+    </GCard>
+</template>
+
+<script setup lang="ts">
+import stores from '@/stores'
+import type { Entity, Term } from '@/types/evaluation'
+import type { Field, TableData } from '@/types/ui/table'
+import * as API from '@/api/evaluation'
+
+const jobSelection = stores.useJobSelection()
+const corporaStore = stores.useCorpora()
+
+const selectedDoc = ref<string>()
+
+const items = ref<Entity[]>()
+const title = computed<string>(() => {
+    return `Entities in ${selectedDoc.value}`
+})
+const columns = ref<Field[]>([
+    { key: "first", label: "NER" },
+    { key: "second" },
+    { key: "third", label: "Count" }
+])
+
+const loading = ref<boolean>(false)
+
+watch(() => selectedDoc.value, () => {
+    loading.value = true
+    API.getEntities(
+        corporaStore.activeUUID,
+        jobSelection.hypothesisJobId,
+        selectedDoc.value
+    ).then(res => {
+        items.value = res.data
+    }).finally(() => {
+        loading.value = false
+    })
+})
+
+function displayEntity(terms: Term[]): string {
+    return terms.map(term => term.annotations["token"]).join(' ')
+}
+</script>

server/src/main/kotlin/org/ivdnt/galahad/web/controller/EvaluationController.kt

@@ -198,5 +198,5 @@ class EvaluationController(
         @PathVariable @Parameter(description = "Corpus UUID") corpus: UUID,
         @PathVariable @Parameter(description = "Document name") document: String,
         @PathVariable @Parameter(description = "Tagger name or sourceLayer") job: String,
-    ): List<Pair<String, List<Term>>> = evaluationService.getEntities(corpus, document, job)
+    ): List<Triple<String, List<Term>, Int>> = evaluationService.getEntities(corpus, document, job)
 }

server/src/main/kotlin/org/ivdnt/galahad/web/service/EvaluationService.kt

@@ -278,7 +278,7 @@ class EvaluationService(val corpora: CorporaService) {
         return cm
     }
 
-    fun getEntities(corpus: UUID, document: String, job: String): List<Pair<String, List<Term>>> {
+    fun getEntities(corpus: UUID, document: String, job: String): List<Triple<String, List<Term>, Int>> {
         val layer = corpora.readAsReaderOrThrow(corpus, user).jobs.readOrThrow(job).getLayer(document)
         return layer.documents.flatMap {
             it.paragraphs.flatMap {
@@ -287,6 +287,8 @@ class EvaluationService(val corpora: CorporaService) {
                         ?.map { span -> span.value to span.indices.map { sent.terms[it] } } ?: emptyList()
                 }
             }
-        }
+        }.groupBy { it }.mapValues { it.value.size }.map { (key, value) ->
+            Triple(key.first, key.second, value)
+        }.sortedByDescending { it.third }
     }
 }