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