Click to edit Master title style UNCLASSIFIED//FOR OFFICIAL USE ONLY Dr. Russell D. Richardson, G2/INSCOM Science Advisor UNCLASSIFIED//FOR OFFICIAL USE ONLY 1
UNCLASSIFIED Semantic Enrichment of the Data Entity Centric Precision Increasing Semantic Richness Solving the Precision / Recall Conundrum with Semantic Enrichment of the Data Concepts/Summarization (e.g. Terrorist Cell Leaders) Resolved Entity (John with ID xxxxx) Entity (Person, Object, Organization, Location) Semantically Labeled Data Po opulation Cen ntric Recall Increasing Anonymity Non- Attributable Aggressively Index Enabled by fine grain security and compliance enforcement De Anonymization of Large Data Sets Detect / Match Behaviors and Patterns Massive Data Sets for Anomaly / Change Detection Massive Data Aggregation for Machine Analytics, Baselining, and Trend Analysis Multi token/lemma/contexual Element/ Part of Speech (Noun, Pronoun, Punctuation) Token (Aggressively Indexed Words) Determine that Two Patterns of Life are the Same but Not Necessarily Whose Pattern of Life Indications and Warnings Non Attributable Aggregate Behavior Determine Avg Traffic Speed by Tracking Cell Movement Determine the Sentiment of a Town, City, Region, Country UNCLASSIFIED 2
Extending Cloud-Enabled Advanced Analytics UNCLASSIFIED All-Source Analytics for Big Data with Advances in Geospatial Indexing, Voice Index and Search, Biometric i Entity Management, Motion Imagery Tracks, Multi-INT Visualization, Collection Management, Support for Mobile Devices, Powerful Compute Platforms, Multi-Level Security, IC Shared Software,. Mobile Voice Motion Imagery Tracks
Click to edit Master title style Fusion Challenges Social Media All Source INTs Person Location Tipping and Cueing Pattern of Life Analysis Alerts & Notifications Information and Situation Sensors Information requests Tactical Reports HUMINT Biometrics DOMEX SIGINT GEOINT Cyber Org Unit ISR Optimization Persistent and Total Entity and Asset Tracking in Entity Database Threat Characterization COA Analysis Continuously & Always Correlating All Data into the Entity Database Challenges 1. Cross INT Correlation 2. Entity Resolution and Disambiguation 3. Scale of the Entity Database 4. Velocity of Data Collection and Processing 5. Dt Determining ii Ptt Patterns of Life and Major Combat Operations with Tolerance to Errors 6. Ranking Threat Severity and Timing 7. Optimizing / Synchronizing ISR 8. Real time Tipping and Cueing 4
Click to edit Need Master to Work title Entity style Tracking Threat Characterization Social Media Reports/ HUMINT Biometrics DOMEX SIGINT GEOINT Cyber Entity Database Person Location Org Unit 4 Optimize ISR ISR Form Links Precorrelat ted Linked Data base P1 P2 P3 P4 P5 L1 L2 L3 L4 O1 O2 O3 O4 O5 Persist Tracks 3 COAs (T2,L3) (T3,L3) (T4,L3) (T1,O1) (T3,L3) (T3,L2) (T1,L2) (T2,L2) (T3,L3) (T1,L1) (T2,L2) (T3,L3) (T1,P1) (T2,P1) (T1,P4) (T3,P3) (T1,P4) (T2,P5) (*,L1) (T1,P1) (T3,L3) 1 2 U1 U2 (T1,L1) (T1,L1) (T2,L1) (T2,L1) (T3,L3) (T3,L3) U3 U4 U5 T1 (P1,L1) (P3,L1) (T3,L3) T2 (P1,L1) (P2,L3) (T3,L3) T3 (P1,L3) (P3,L3) (T3,L3) T4 (P2,L3) (T2,L1) (T3,L3) T5 Same Time, Same Place L2 may be an important location as many P s have been reported being there Patterns of Life Determined P1 and P3 share location pattern Infer P3 at L1 as O1 is at L1 Location of P1 over time 5
The DCGS-A ICITE Cloud (aka Red Disk) Click to edit Master title style at-scale Entity Database Sensors Sources SIGINT MTI WAMI FMV TED Fires Harmony USMFT Collection Mgmt Open Source Mission Command Audio etc Velocity & Content EDH Provenance TDF Metadata tagging Geo temporal extraction Entity extraction and nomination Artifact enrichment Security Labeling Metrics more RTAAP NIFI / Storm Real Time Advanced Analytics Pipeline updating analytics earliest as possible. Data Interoperability Decreases Information Enterprise Context Burden Full Spectrum Analytic Awareness Artifacts, Terms Statements Analytics and indexes updated wrt to each artifact Maximally correlated data Contextual based navigation All data under a common representation to enable assess to all layers Cross corpus analytics without custom code Signatures are analyzed in real time Logical representation needs to be consistent in order for the physical model and view to be dynamic M/R MR Data Access Process User s authorizations and roles are matched to data security labels UCD Enrichment Analytics and indexes wrt to the entire corpus, Bulkand incremental updates Data sharing Analyst s conclusions enrich the UCD DIAS User s Authorizations Community Partners Classes of relationships determined at various points No all relationships need to be explicitly expressed correlations enable this All analytic disciplines together with correlated data among all disciplines Analytic models are represented in the UCD 6