feat(impression): introduce active entity recall model

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/ActiveEntity.kt

@@ -3,18 +3,30 @@ package work.slhaf.partner.core.cognition.impression
 import org.w3c.dom.Document
 import org.w3c.dom.Element
 import work.slhaf.partner.core.cognition.context.BlockContent
+import java.time.Instant
+import java.time.ZoneId
+import java.util.*
 import java.util.concurrent.atomic.AtomicReference
 
 class ActiveEntity @JvmOverloads constructor(
-    timestamp: Long = System.currentTimeMillis(),
+    val runtimeId: String = newActiveEntityRuntimeId(),
-    private val _evidences: MutableList<String> = mutableListOf(),
+    val createdAt: Instant = Instant.now(),
-) : BlockContent("active_entity_$timestamp", "impression") {
+    boundEntityUuid: String? = null,
-    val evidences: List<String>
+    private val _evidences: MutableList<EntityEvidence> = mutableListOf(),
+) : BlockContent("active_entity_$runtimeId", "impression") {
+    val evidences: List<EntityEvidence>
         get() = synchronized(_evidences) { _evidences.toList() }
 
+    @Volatile
+    var lastMentionedAt: Instant = createdAt
+        private set
+
     private val _subject = AtomicReference("UNKNOWN")
     val subject: String get() = _subject.get()
 
+    private val _boundEntityUuid = AtomicReference<String?>(boundEntityUuid)
+    val boundEntityUuid: String? get() = _boundEntityUuid.get()
+
     private val _projectedFeatures: MutableMap<String, Double> = mutableMapOf()
     val projectedFeatures: Map<String, Double>
         get() = synchronized(_projectedFeatures) { _projectedFeatures.toMap() }
@@ -23,12 +35,27 @@ class ActiveEntity @JvmOverloads constructor(
     val projectedImpressions: Map<String, Double>
         get() = synchronized(_projectedImpressions) { _projectedImpressions.toMap() }
 
-    fun addEvidence(evidence: String) = synchronized(_evidences) {
+    @JvmOverloads
+    fun addEvidence(
+        content: String,
+        associationConfidence: Double = 1.0,
+        source: EntityEvidence.Source = EntityEvidence.Source.USER_INPUT,
+        timestamp: Long = System.currentTimeMillis(),
+    ) = addEvidence(EntityEvidence(content, associationConfidence, source, timestamp))
+
+    fun addEvidence(evidence: EntityEvidence) = synchronized(_evidences) {
         _evidences.add(evidence)
+        touch(Instant.ofEpochMilli(evidence.timestamp))
     }
 
     fun updateSubject(subject: String) = _subject.set(subject)
 
+    fun bindEntity(uuid: String?) = _boundEntityUuid.set(uuid)
+
+    fun touch(time: Instant = Instant.now()) {
+        lastMentionedAt = time
+    }
+
     fun addProjectedFeatures(vararg features: Pair<String, Double>) = synchronized(_projectedFeatures) {
         features.forEach { _projectedFeatures[it.first] = it.second }
     }
@@ -38,6 +65,11 @@ class ActiveEntity @JvmOverloads constructor(
     }
 
     override fun fillXml(document: Document, root: Element) {
+        root.setAttribute("runtime_id", runtimeId)
+        boundEntityUuid?.let { root.setAttribute("bound_entity_uuid", it) }
+        root.setAttribute("created_at", modelTime(createdAt))
+        root.setAttribute("last_mentioned_at", modelTime(lastMentionedAt))
+
         appendTextElement(document, root, "subject", subject)
 
         appendListElement(
@@ -46,7 +78,14 @@ class ActiveEntity @JvmOverloads constructor(
             "evidences",
             "evidence",
             synchronized(_evidences) { _evidences.toList() }
-        )
+        ) { evidence ->
+            setAttribute("association_confidence", evidence.associationConfidence.toString())
+            setAttribute("source", evidence.source.name)
+            setAttribute("timestamp", evidence.timestamp.toString())
+            setAttribute("truncated", evidence.isContentTruncated().toString())
+            setAttribute("original_length", evidence.content.length.toString())
+            textContent = evidence.contentForContext()
+        }
 
         appendListElement(
             document,
@@ -70,4 +109,10 @@ class ActiveEntity @JvmOverloads constructor(
             textContent = entry.key
         }
     }
+
+    private fun modelTime(time: Instant): String =
+        time.atZone(ZoneId.systemDefault()).toString()
 }
+
+private fun newActiveEntityRuntimeId(): String =
+    UUID.randomUUID().toString().replace("-", "").take(12)

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/EntityEvidence.kt

@@ -0,0 +1,112 @@
+package work.slhaf.partner.core.cognition.impression
+
+/**
+ * Runtime evidence associated with an active entity.
+ *
+ * The confidence describes how strongly this evidence is associated with the
+ * current active entity, not whether the evidence content itself is true.
+ */
+data class EntityEvidence @JvmOverloads constructor(
+    val content: String,
+    val associationConfidence: Double = 1.0,
+    val source: Source = Source.USER_INPUT,
+    val timestamp: Long = System.currentTimeMillis(),
+) {
+    enum class Source {
+        USER_INPUT,
+        ASSISTANT_REPLY
+    }
+
+    fun isContentTruncated(maxLength: Int = CONTEXT_CONTENT_MAX_LENGTH): Boolean =
+        content.length > maxLength
+
+    fun contentForContext(maxLength: Int = CONTEXT_CONTENT_MAX_LENGTH): String {
+        if (content.length <= maxLength) {
+            return content
+        }
+
+        val available = maxLength - OMITTED_MARKER.length
+        if (available <= 0) {
+            return content.take(maxLength)
+        }
+
+        val headBudget = available / 2
+        val tailBudget = available - headBudget
+        val headEnd = adjustHeadEnd(content, headBudget)
+        val tailStart = adjustTailStart(content, content.length - tailBudget)
+
+        if (tailStart <= headEnd) {
+            return content.take(maxLength).trimEnd()
+        }
+
+        return content.substring(0, headEnd).trimEnd() +
+                OMITTED_MARKER +
+                content.substring(tailStart).trimStart()
+    }
+
+    private fun adjustHeadEnd(source: String, preferredEnd: Int): Int {
+        val safePreferredEnd = preferredEnd.coerceIn(0, source.length)
+        findForwardBoundary(source, safePreferredEnd, STRONG_BOUNDARY_SEARCH_WINDOW, ::isStrongBoundary)?.let {
+            return it + 1
+        }
+        findForwardBoundary(source, safePreferredEnd, SOFT_BOUNDARY_SEARCH_WINDOW, ::isSoftBoundary)?.let {
+            return it + 1
+        }
+        return safePreferredEnd
+    }
+
+    private fun adjustTailStart(source: String, preferredStart: Int): Int {
+        val safePreferredStart = preferredStart.coerceIn(0, source.length)
+        findBackwardBoundary(source, safePreferredStart, STRONG_BOUNDARY_SEARCH_WINDOW, ::isStrongBoundary)?.let {
+            return it
+        }
+        findBackwardBoundary(source, safePreferredStart, SOFT_BOUNDARY_SEARCH_WINDOW, ::isSoftBoundary)?.let {
+            return it
+        }
+        return safePreferredStart
+    }
+
+    private fun findForwardBoundary(
+        source: String,
+        start: Int,
+        window: Int,
+        predicate: (Char) -> Boolean,
+    ): Int? {
+        val end = (start + window).coerceAtMost(source.length)
+        for (index in start until end) {
+            if (predicate(source[index])) {
+                return index
+            }
+        }
+        return null
+    }
+
+    private fun findBackwardBoundary(
+        source: String,
+        start: Int,
+        window: Int,
+        predicate: (Char) -> Boolean,
+    ): Int? {
+        val end = (start - window).coerceAtLeast(0)
+        for (index in start downTo end + 1) {
+            if (predicate(source[index - 1])) {
+                return index
+            }
+        }
+        return null
+    }
+
+    private fun isStrongBoundary(char: Char): Boolean = char == '\n'
+
+    private fun isSoftBoundary(char: Char): Boolean = when (char) {
+        '。', '！', '？', ';', '；', '.' -> true
+        else -> false
+    }
+
+    companion object {
+        const val CONTEXT_CONTENT_MAX_LENGTH = 480
+        private const val OMITTED_MARKER = "\n...[omitted]...\n"
+        private const val STRONG_BOUNDARY_SEARCH_WINDOW = 120
+        private const val SOFT_BOUNDARY_SEARCH_WINDOW = 80
+    }
+}

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/ImpressionVectorIndex.java

@@ -22,7 +22,7 @@ public class ImpressionVectorIndex {
     }
 
     public void upsert(String text, Entity.IndexableData indexableData){
-        if (VectorClient.status){
+        if (!VectorClient.status){
             return;
         }
         String modelId = VectorClient.VECTOR_MODEL_ID;

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/EntityAssociationMatch.kt

@@ -0,0 +1,7 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+data class EntityAssociationMatch(
+    val target: ImpressionSearchTarget,
+    val score: Double,
+    val hits: List<ImpressionSearchHit> = emptyList(),
+)

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionSearchDocument.kt

@@ -0,0 +1,10 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+data class ImpressionSearchDocument(
+    val id: String,
+    val target: ImpressionSearchTarget,
+    val field: ImpressionSearchField,
+    val text: String,
+    val weight: Double = 1.0,
+    val metadata: Map<String, String> = emptyMap(),
+)

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionSearchDocuments.kt

@@ -0,0 +1,138 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+import work.slhaf.partner.core.cognition.impression.ActiveEntity
+import work.slhaf.partner.core.cognition.impression.Entity
+
+object ImpressionSearchDocuments {
+
+    fun fromActiveEntity(activeEntity: ActiveEntity): List<ImpressionSearchDocument> {
+        val target = ImpressionSearchTarget(
+            ImpressionSearchTarget.Type.ACTIVE_ENTITY,
+            activeEntity.runtimeId
+        )
+        val metadata = activeEntity.boundEntityUuid
+            ?.let { mapOf("boundEntityUuid" to it) }
+            .orEmpty()
+
+        return buildList {
+            add(
+                ImpressionSearchDocument(
+                    id = "active:${activeEntity.runtimeId}:subject",
+                    target = target,
+                    field = ImpressionSearchField.SUBJECT,
+                    text = activeEntity.subject,
+                    weight = SUBJECT_WEIGHT,
+                    metadata = metadata,
+                )
+            )
+
+            activeEntity.evidences.forEachIndexed { index, evidence ->
+                add(
+                    ImpressionSearchDocument(
+                        id = "active:${activeEntity.runtimeId}:evidence:$index",
+                        target = target,
+                        field = ImpressionSearchField.EVIDENCE,
+                        text = evidence.contentForContext(),
+                        weight = EVIDENCE_WEIGHT * evidence.associationConfidence,
+                        metadata = metadata,
+                    )
+                )
+            }
+
+            activeEntity.projectedFeatures.entries.forEachIndexed { index, entry ->
+                add(
+                    ImpressionSearchDocument(
+                        id = "active:${activeEntity.runtimeId}:feature:$index",
+                        target = target,
+                        field = ImpressionSearchField.FEATURE,
+                        text = entry.key,
+                        weight = FEATURE_WEIGHT * entry.value,
+                        metadata = metadata,
+                    )
+                )
+            }
+
+            activeEntity.projectedImpressions.entries.forEachIndexed { index, entry ->
+                add(
+                    ImpressionSearchDocument(
+                        id = "active:${activeEntity.runtimeId}:impression:$index",
+                        target = target,
+                        field = ImpressionSearchField.IMPRESSION,
+                        text = entry.key,
+                        weight = IMPRESSION_WEIGHT * entry.value,
+                        metadata = metadata,
+                    )
+                )
+            }
+        }
+    }
+
+    fun fromEntity(entity: Entity): List<ImpressionSearchDocument> {
+        val target = ImpressionSearchTarget(
+            ImpressionSearchTarget.Type.ENTITY,
+            entity.uuid
+        )
+
+        return buildList {
+            add(
+                ImpressionSearchDocument(
+                    id = "entity:${entity.uuid}:subject",
+                    target = target,
+                    field = ImpressionSearchField.SUBJECT,
+                    text = entity.subject,
+                    weight = SUBJECT_WEIGHT,
+                )
+            )
+
+            entity.snapshotFeatures().keys.forEachIndexed { index, feature ->
+                add(
+                    ImpressionSearchDocument(
+                        id = "entity:${entity.uuid}:feature:$index",
+                        target = target,
+                        field = ImpressionSearchField.FEATURE,
+                        text = feature,
+                        weight = FEATURE_WEIGHT,
+                    )
+                )
+            }
+
+            entity.snapshotImpressions().keys.forEachIndexed { index, impression ->
+                add(
+                    ImpressionSearchDocument(
+                        id = "entity:${entity.uuid}:impression:$index",
+                        target = target,
+                        field = ImpressionSearchField.IMPRESSION,
+                        text = impression,
+                        weight = IMPRESSION_WEIGHT,
+                    )
+                )
+            }
+
+            entity.showRelations().forEachIndexed { index, relation ->
+                val relationText = buildString {
+                    append(relation.target)
+                    relation.relations.keys.forEach { name ->
+                        append(' ')
+                        append(name)
+                    }
+                }
+
+                add(
+                    ImpressionSearchDocument(
+                        id = "entity:${entity.uuid}:relation:$index",
+                        target = target,
+                        field = ImpressionSearchField.RELATION,
+                        text = relationText,
+                        weight = RELATION_WEIGHT,
+                    )
+                )
+            }
+        }
+    }
+
+    private const val SUBJECT_WEIGHT = 1.0
+    private const val FEATURE_WEIGHT = 0.85
+    private const val IMPRESSION_WEIGHT = 0.75
+    private const val RELATION_WEIGHT = 0.65
+    private const val EVIDENCE_WEIGHT = 0.8
+}

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionSearchField.kt

@@ -0,0 +1,9 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+enum class ImpressionSearchField {
+    SUBJECT,
+    FEATURE,
+    IMPRESSION,
+    RELATION,
+    EVIDENCE
+}

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionSearchHit.kt

@@ -0,0 +1,7 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+data class ImpressionSearchHit(
+    val document: ImpressionSearchDocument,
+    val score: Double,
+    val matchedTerms: Set<String> = emptySet(),
+)

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionSearchTarget.kt

@@ -0,0 +1,11 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+data class ImpressionSearchTarget(
+    val type: Type,
+    val id: String,
+) {
+    enum class Type {
+        ACTIVE_ENTITY,
+        ENTITY
+    }
+}

Partner-Core/src/main/java/work/slhaf/partner/core/cognition/impression/search/ImpressionTextSearch.kt

@@ -0,0 +1,15 @@
+package work.slhaf.partner.core.cognition.impression.search
+
+interface ImpressionTextSearch {
+
+    fun rebuild(documents: Collection<ImpressionSearchDocument>)
+
+    fun upsert(document: ImpressionSearchDocument)
+
+    fun removeByTarget(target: ImpressionSearchTarget)
+
+    fun search(
+        query: String,
+        limit: Int = 20,
+    ): List<ImpressionSearchHit>
+}