The 10th International Conference on Big Data and Information Analytics (BigDIA 2024)

A Brief Analysis of Palantir Gotham: A

Collaborative and Interactive Big Data Visualization
Analysis Software Based on Dynamic Ontology

Wei Wei Xiang Li Kun Liu Siyu Tan *

Applied Search Dept. Applied Search Dept. Applied Search Dept. Colleage of Systems
Advanced Insititute of Big Data Advanced Insititute of Big Data Advanced Insititute of Big Data Engineering
Beijing, China Beijing, China Beijing, China National University of Defense
[email protected] [email protected] [email protected] Technology
Changsha, China
2024 10th International Conference on Big Data and Information Analytics (BigDIA) | 979-8-3503-5462-1/24/$31.00 ©2024 IEEE | DOI: 10.1109/BIGDIA63733.2024.10808897

[email protected]

Abstract—Palantir Gotham is a big data analysis and for business analysts within Gotham. Employing interface
management software based on dynamic ontology technology, analysis and process analysis methodologies, it organizes the
catering to the domains of military decision-making, national stages of collaborative interactive analysis of big data
defense security, government administration, etc. This paper selects featuring multi-source heterogeneity, multi-modality, and
an frontend analytics software targeted at business analysts from multiple AI models based on this software. Key capabilities
Gotham, delves into its primary capabilities and core and functions are analyzed and outlined. Subsequently, the
functionalities, examines and organizes the underlying core paper summarizes Gotham's core technologies, namely,
technologies, primarily consisting of big data integration and dynamic ontology-based big data integration and business
business mapping techniques based on dynamic ontologies, an
mapping technology, integrated secure collaboration-oriented
integrated security-enhanced collaborative big data system
technical architecture, and human-centered collaborative
big data system technical architecture, and human-centric
interactive visualization analysis techniques for big data. The paper interactive visualization analytics for big data.
analyzes and summarizes the primary mapping principles of Within this context, through document analysis, patent
structured data within the Palantir ontology, the key application analysis, and architecture analysis, the paper explains the
principles of Palantir's dynamic ontology technology, along with its primary mapping principles of structured data within the
comprehension of the three-layer structure and three characteristics Palantir Ontology, contrasting traditional data manipulation
of the dynamic ontology. The paper outlines and summarizes the
with Palantir's ontology-based Action Type approach. The
understanding diagram of Palantir's big data system technical
paper summarizes the main operational principles of Palantir's
architecture, identifying and elucidating its crucial technological
components, and identifies and summarizes three typical frontend dynamic ontology technology, expounds upon the three-layer
styles employed in Gotham's interactive visualization analysis. understanding of its dynamic ontology, and proposes three
Ultimately, the paper shares three insights: human-computer characteristics of this dynamic ontology.
collaborative analysis, the philosophy of dynamic ontology, and Given the lack of publicly available information on
collaborative interactive analysis. Palantir's big data system technical architecture, the paper
compiles and summarizes a comprehensive understanding
Keywords—Big Data Analysis, Collaborative Interactive
Visualization, Dynamic Ontology, Palantir Gotham, Human-
diagram of Palantir's big data system architecture through data
Machine Collaboration collection and analysis, identifying and elucidating its key
technological components. Furthermore, through software
I. INTRODUCTION interface analysis and human-computer interaction analysis,
the paper identifies and summarizes three typical front-end
Palantir's big data analytics technology grounded in styles employed in Gotham's interactive visualization
dynamic ontologies, alongside its collaborative and analysis.
interactive analysis software supporting distributed geo-
environments, has achieved remarkable success across Finally, the paper shares insights gained in three areas:
military, defense, security, and commercial sectors, garnering human-machine collaborative analysis, the philosophy of
recognition from users in the military, government, and dynamic ontology, and collaborative interactive analysis.
various industries such as finance. Palantir's digital twin
modeling paradigm, which integrates data and models based II. OVERVIEW OF PALANTIR
on dynamic ontologies, coupled with its intuitive, wizard- A. Introduction to Palantir Technologies Inc.
driven modeling software tools and collaborative interactive
analysis tools, enables business users to bypass the constraints Palantir Technologies Inc. (Palantir for short) [1], founded
of complex programming to a certain extent, enabling them to in the United States in 2004, is a digital technology service
dynamically and swiftly leverage data for decision-making. provider and a big data technology company. Its clientele
encompasses military and government agencies, investment
This paper, after introducing Palantir Inc. and its product banks, etc. Palantir has long been dedicated to and excelled in
Palantir Gotham, selects a front-end analysis software tailored fusing dispersed, multi-source, and heterogeneous data into
*Corresponding author

979-8-3503-5462-1/24/$31.00 ©2024 IEEE

769
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
high-value digital assets based on dynamic ontologies, as well B. Analysis of the Software’s Display Interface
as integrating multi-modal data into business processes, The software is tailored for business analysts, based on
thereby enabling clients to achieve decision-making customers' actual business processing logic and workflows, by
operational systems grounded in big data. Originating from providing customized analytical applications, data
developing customized data analysis systems for the CIA [2], presentation and interaction methods, as well as interlinking
Palantir has evolved to encompass multiple platform products mechanisms for various functions and controls. It dynamically
such as Apollo, Gotham, Foundry, and AIP [3][4]. presents prediction, simulation, and emulation processes,
FOUNDRY Active LLMs on private
facilitating business analysts to intuitively explore and analyze
A GOTHAM
network AIP CORE information and data in a visually appealing manner. This
P SaaS design fosters excellent cooperation, drive, and inspiration
AIP ACTION
Continuous Delivery
Automated Upgrades O Govern and Defining
different LLMs GRAPH among users.
Automated Configuration GOTHAM
L Connected
Data sensitivity, Laws
and regulations 2
Appliance AIP 3
L Monitor & control LLM CONTROL
GOTHAM Activity
Monitor from a single
O Disconnected Control, Govern and
PLANE
Pane of glass Operations Trust AI 4 5

Fig. 1. The Four Major Platform of Palantir [Source: Palantir Blog, Dolphin
Research]

B. Introduction to Palantir Gotham 1

Palantir Gotham [5] (Gotham for short) is a big data
analysis platform with a series of tools that fosters a symbiotic 6
relationship between human brain decision-making and
computer intelligence [6]. It serves the fields of military
decision-making, national defense security, and government Fig. 3. Typical Display Interface of Gotham[Source: Palantir]
administration. Highly customized to meet client needs,
Gotham provides analysts with a comprehensive suite of Fig.3 presents a representative screenshot of the software
collaborative and interactive visualization analysis tools, display interface of Gotham:
supporting the semantic, temporal, and geospatial correlation
Area ①: the primary function area contains various icon
and causal analysis of both structured, notably, unstructured
data. groups tailored to customer business logic, such as Situation
(based on GAIA), Video, Report, Table, Mission, Graph and
III. ANALYSIS OF PALANTIR GOTHAM ANALYSIS SOFTWARE other applications. These applications are universal within the
domain, abstracted and modeled by deployment strategists
The paper, based on the frontend analysis based on the business logic, core processes, and key activities
software(hereinafter referred to as “the Software”) in the customer's business area.
demonstrated in the Palantir Gotham official demonstration
video, explores the functionality and technical support of Area ② : features like global search, browser, online
Gotham in enabling collaborative and interactive visualization discussion, and file sharing are provided, supporting semantic
analysis of integrated multi-source heterogeneous big data search, instant messaging, and data resource sharing.
from a user's perspective.
The central area comprises multiple components and is the
A. Collaborative and Interactive Analysis Process core functional area. Area ③: includes a menu bar, status
Using notional scenarios and data, with a backdrop of a information, and online collaboration tools. Area ④ : the
specific situation, the demonstration video showcases how panels on the left/right sides display object details, enabling
business analysts utilize the Software to conduct collaborative data querying and interactive operations. Area ⑤: serves as
and interactive visualization analysis in the following five the visualization and interaction area for information and data,
stages: supporting 3D model manipulation, detailed information
Phase 1
viewing, and various analytical operations. Area ⑥: further
Internal Situation exhibits linked details and supports timeline control.
Information Sharing and
Interactive Viewing Additionally, the interface incorporates display areas for
Phase 4 processing results, alerts, and process information. By
Action suggestions and
Phase 2 simulation analysis utilizing pop-up bubbles and notifications, it promptly
External Alert communicates status changes and alerts, allowing for further
Information Sharing and
Interactive Viewing
actions, thereby enhancing the software's interactivity and
Phase 5 practicality.
Continuous monitoring
Phase 3 of action process C. Analysis of Key Capabilities and Functions of the
Target Movement Software
Prediction and Real-time
Reconnaissance & Based on the interactive analysis process for multi-source
Analysis and multi-modal big data in scenarios and the static analysis
results of the software's main display interface, the primary
Fig. 2. The main stages of interactive analysis in the video scenarios

770
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
capabilities and functions of this software are summarized as In the Palantir platform, the definitions of Objects, Links,
follows: and Actions collectively constitute the so-called ontology—a
digital representation of the organization [8]. Objects can be
1) Key Capabilities categorized into regular entity objects and two types of special
• Interactive Analysis and Online Collaboration entity objects: Document and Event. Specifically, Document
• Real-time Data Integration, Sharing, and Computation Objects primarily refer to unstructured textual data content
itself, while Event Objects signify occurrences based on time.
• AI Integration and Capabilities
The basic mapping principles for structured data within the
• Dynamic Ontology-Related Capabilities Palantir ontology are illustrated in the diagram below:
2) Main Functions Datasets Palantir Ontology UML Element
• Interactive Analysis, Online Collaborative Analysis,
Online Approval Dataset Object Type Class

• Real-time Data Integration and Sharing Row Object Object

Column Property Attribute

• AI Model Integration, AI Capabilities(Real-time
Analysis, Recommendation, Prediction, NLP) Field Property Value Attribute Value

• Ontology Extraction, Object-based Information Join Link Type Relationship

Association
table_shipinfo Object Set
Object Type
• Real-time Computation, Geo-grid-based Computation, ship_id ship_name ship_length Object Object
Simulation Integration, Online Simulation Ship Ship_Pacific ...
ID S001 ...
• Task Storage, Investigation Storage S0001 Pacific 140 Name Pacific ...

... ... ... Length 140 ...

IV. ANALYSIS OF CORE TECHNOLOIES
This subsection delves into the analysis of the core
technologies employed within the primary capabilities and Fig. 5. Structured datasets-Ontology Mapping
functionalities of the software.
Below is the example of an ontology object presented in
A. Dynamic-ontology-based Big Data Organization, Palantir Object View:
Fusion, and Business Mapping Technology
Ontology is more than just a tool; it maps the business
processes of Palantir's clients. Palantir's dynamic ontology
technology transforms clients' digital assets into a dynamic,
operable foundation [7].
1) Introduction to Palantir Ontology
"Ontology" refers to the semantic description of a domain-
specific set of normative concepts and their inherent logical
relationships. Its role is to provide a fundamental classification
framework and association description mechanism for domain
data information, thereby enabling the rational organization of
the semantics embedded within the data.

Fig. 6. Intuitively Creating Action Types with Palantir Ontology Manager

[Source: Palantir]

In terms of handling data operations, traditional methods

involve discrete and direct manipulations of data in database
tables through hardcoded logic, requiring users to possess
programming skills. Moreover, the business logic and
operational processes are not intuitive. Conversely, Palantir's
ontology employs the core concepts of Objects and Actions,
offering an Ontology Manager that facilitates a guided and
intuitive approach to modeling. This allows users to quickly
and conveniently manipulate data organized by Objects based
on business logic, as illustrated in the Fig.7:

Fig. 4. Ontologies in Palantir Platform Software [Source: Palantir]

771
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
 in ontology modeling. The fundamental principle of utilizing
dynamic ontology in Gotham is rooted in its ability to
integrate diverse and heterogeneous data, transforming them
into the data object format utilized within Gotham. This
process can adapt to changes in clients' business models and
data sources, significantly streamlining the data integration
process. And by connecting underlying data and models to
their real-world business objects, Gotham allows users to
collaborate on dynamic knowledge assets in real-time [14].

Fig. 7. Intuitively Creating Action Types with Palantir Ontology Manager

[Source: Palantir]

Fig.8 shows a comparative illustration of traditional data

operations and data operations based on Palantir Ontology
Action Type:

[Prerequisite 1] [Step 1]
Should master SQL or other Define Action Type Change Name, Fig. 9. Palantir Dynamic Ontology[Source: Palantir]
programming skill Select Change object(s),chooes
[Prerequisite 2] Object Pacific and Modify property
Should master the underlying Name The Palantir Dynamic Ontology encompasses three layers
database tables
[Step 2]
of meaning:
AND type following code: Do the Action
UPDATE Table_shipinfo (1) Semantic Layer: its core components primarily consist
SET ship_name = 'Arctic'
Manger Ship
of dynamic objects & links, multi-modal attributes, and
WHERE Ship_id = 'S0001';
Amanda Pacific
ontology primitives. In this layer, by mapping existing data
table_shipinfo Ship Ship_Pacific
sources into ontology Objects, Object Properties, and Links
Change
ship_id ship_name ship_length Name ID S001
between Objects, it establishes a semantic foundation for the
Pacific Pacific Kinetic Layer. This foundation includes comprehensive
S0001 140 Name
Arcitc Arcitc metadata, fine-grained security controls, and governance over
... ... ... Length 140
all changes.
Fig. 8. Traditional Data Manipulation versus Ontology-based Data (2) Kinetic Layer: its core functionalities encompass AI-
Manipulation with ActionType driven actions and functions, process mining & automation,
action orchestration, and real-time monitoring. Building upon
As can be seen from Fig.8, traditional data operations the Semantic Layer, this layer leverages the Action Types
necessitate collaboration between business personnel and within the ontology to capture data during various operations,
developers, where developers must possess relevant orchestrate decision-making processes connected to existing
programming skills and an intimate understanding of the systems, and model organizational kinetics. Additionally, it
detailed definitions of database tables. This approach suffers provides Functions as a means to implement arbitrarily
from requiring a large number of personnel, lengthy complex business processing logic.
development timelines, high technical barriers, and an
obscured presentation of business logic. In contrast, by (3) Dynamic Layer: its core capabilities include AI-
utilizing Palantir's ontology-based Action Type for data powered decision-making, multi-step simulations, and
operations, business personnel can leverage a guided frontend decision capture & learning. Drawing from the core
software tool to select pre-defined ontology Objects based on components and functionalities of both the Semantic and
their own business processing logic. This approach requires Kinetic layers, this layer deeply integrates the ontology into
fewer personnel, simplifies operation definitions, lowers the user-facing front-end analysis and operational software,
technical threshold, and clearly presents business logic. enabling dynamic decision-making within human-computer
interactions.
2) Palantir Dynamic Ontology Technology
Dynamic ontology typically refers to the capability of This paper argues that the Palantir Dynamic Ontology
flexibly expanding, modifying, or deleting ontology content primarily exhibits the following three characteristics:
during system operation [9]. It enables automatic updates and (1) Highly Universal Meta-ontology Concepts and Highly
accomplishes data integration when data changes or new data Customizable Domain-specific Ontology Content: It
sources are added [10]. This is manifested primarily in aspects establishes top-level meta-ontology concepts such as "Entity,"
such as self-adjustability, dynamic generation, dynamic "Event," and "Document," which possess cross-domain
adjustment of models, and dynamic evolution of knowledge. universal applicability. Additionally, users can customize
Traditional ontology modeling software [11][12][13] ontology content relevant to their specific domains.
struggle to effectively support scenarios with frequent (2) Dynamic Maintenance and Updates to Maintain the
changes in requirements, especially when there are high Timeliness of Ontology Knowledge Information: The
demands for agility, timeliness, completeness, and ease of use

772
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
dynamic ontology technology supports flexible modifications Fig.11 shows the example of Lineage in Palantir,
to existing ontologies, ensuring the timeliness of ontology illustrating the detailed lineage from raw data sourced from
knowledge information. It also enables auxiliary multiple structured database tables to the object entities within
modifications to existing objects, attributes, and relationships, the ontology [17]. This demonstration underscores how data
such as adding and modifying tags, icons, parsers, among is progressively reconstituted to align with the client's
others. objective world and business perspectives in terms of data
organization.
(3) Specialized Parsers for Attribute Types Facilitate
Heterogeneous Data Integration: It allows users to define
dynamic ontologies named "database." Through parsers,
multi-source heterogeneous data can be integrated, mapping
input data to the dynamic ontology. This feature simplifies the
integration process and enhances data interoperability.
B. Big Data System Technical Architecture for Integrated,
Secure, Collaboration and Related Technologies
1) Technical Architecture of Big Data
Through analyzing and organizing the software tools and
controls mentioned in Palantir's public materials [15][16], this
paper forms an understanding of its technical architecture for
big data systems:

Gotham- Foundry-
Application GAIA, Dossier, Video,
Monocle,Contour, Fusion, Object AIP
Explorer, Workshop, Vertex,
Layer Brifings, Graph, Table,
AVA, Chat, ...
Code Authoring, Quiver, AI/ML,
AI Product
Fig. 11. An Example of Lineage in Palantir [Source: Palantir]
Code Workshop, Reports, ...

Open Data Pluggable UI/Tools Explanation: In Fig.11, from left to right, the yellow
Analytic Import/Export Platform
Layer elements represent raw data, the green elements represent
Analytic API Palantir Web Palantir Workshop
cleaned data, the orange elements represent transformed data,
Persistent Faceted Search Unstructured the pink elements represent data processed through
Search Search Search Around Search parsing/NLP/ML, and the dark brown elements represent
Layer Search APIs REST WS ontologies.
Horizion
powered

3) Privacy and Security Controls

AIP

Cross Data Source

Revisioning Investigation
The Access Control Interface within Palantir's big data
Resolution
Knowledge Database Repository technical architecture provides built-in role-based,
Lineage Engine
Layer classification-based, and purpose-based access controls to
Dynamic Ontology ensure that authorized users access only the data they are
permitted to view, offering fine-grained privacy and security
Access Control Interface
Security controls for clients. "For highly sensitive data, markings serve
Built-in Role Purpose-based Classification as another form of mandatory controls" [18]. The ontology-
Layer Data Subject Access Right-to-be-forgetten
based data organization introduces a new dimension and
Request Module Module
granularity to data access control, enabling controls to be
Delivery applied specifically to ontology objects, entities, and their
Apollo properties. Additionally, Palantir employs two modules, the
Layer
"Data Subject Access Request Module" and the "Right-to-be
Fig. 10. Understanding Diagram of Palantir's Big Data System Technical -forgotten Module," to assist clients in implementing Articles
Architecture 15 and 16 of GDPR [19], enhancing compliance with data
protection regulations.
The following section provides an brief analysis and
introduction of the key components illustrated in Fig.10. 4) Collaboration and Version Management
The Gotham platform supports collaborative analysis
2) Integration and Ontology Construction across geographically distributed multi-seats, leveraging an
Within the technical architecture of the Palantir big data audit and version control framework. It employs a
system, over 200 data connectors can be utilized. These Revisioning Database and the Nexus Peering Technology of
connectors employ the CEDT framework (Crawl, Extract, the Palantir Gotham platform, enabling multiple users within
Detect, Transform) to collect, clean, quality-check, and a customer organization or across multiple customers to
transform multi-source heterogeneous data. Subsequently, seamlessly and securely conduct collaborative analysis on the
leveraging dynamic ontology technology, ontology same data [20]. Palantir's big data technical architecture
construction is carried out, whereby information and data supports business analysts working in their individual data
fragments pertaining to physical world entities scattered spaces (preserving historical branch records of each revision),
across various data resources are aggregated into object with submissions only being published to the public data space
entities within the ontology. This process reconfigures data upon completion of analysis. The Nexus Peering technology
organization and management, enabling data to be organized achieves data sharing of Palantir Gotham object instances by
and utilized in a manner that aligns more closely with business capturing, cycling, and fusing data changes.
perspectives and operations.

773
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
C. Human-centered Big Data Interactive Visualization and 2) Time-based Information Comparison
Analysis Technologies The software facilitates information comparison across
In section " Analysis of the Software’s Display Interface" different time points by providing a Timeline panel that
showcases fundamental interactive controls, where the display interacts with the main application interface. It also assists
interface design and the interconnectivity between controls, in with AI model recognition functionality for differential
conjunction with application designs that align with clients' annotation or directly presents AI model recognition results
business processing logic, provide users with convenient and comparison effects, as depicted in Fig.12-2.
heuristic and interactive analysis tools. Palantir is also 3) Space-based Query and Calculation
continuously engaged in research on how machines can better The software supports functions such as Collect satellite
assist and drive human operations. The samples of its patents imagery, Geosearch[22], and Create geofence[23], offering
related to human-computer interaction and visualization are business analysts intuitive and visual ways to access and view
presented in Table I and Table II [21]: information, as illustrated in F Fig.12-3.
TABLE I. SAMPLES OF PALANTIR'S PATENTS RELATED TO HUMAN-
COMPUTER INTERACTION AND VISUALIZATION [21]

No. Subhead
1 Systems and methods for visual definition of data associations

Systems and user interfaces for dynamic and interactive access

2 of, investigation of, and analysis of data objects stored in one
or more databases
Systems and user interfaces for holistic, data-driven
3 investigation of bad actor behavior based on clustering and
scoring of related data
Systems and user interfaces for holistic, data-driven
4 investigation of bad actor behavior based on automatic 1 2
clustering of related data in various data structures
5 Relationship visualizations 3
6 System and method for sharing investigation results

7 Systems and methods for visual definition of data associations

TABLE II. SAMPLES OF TOPIC MODELING OF PALANTIR PATENTS [21]

Topic Example

Selecting "Another type of user input device is cursor Fig. 12. Typical Frontend Pattern of Interactive Analysis in the Software
Customizable data control, such as a mouse, a trackball, or [Source: Palantir]
objects in graphs with cursor direction keys for communicating
nodes and edges direction information and command."
The collaborative interaction of this software is primarily
achieved through real-time online sharing of the current
Segmentation and "Rendering one or more graphical data
partitioning objects display widgets according to each display
interface and associated data, online sharing of Dossiers, as
using sets, claims, order value, which widget type and the well as collaboration facilitated by functions such as online
time series, and values widget configuration values" discussions and file sharing in Section III-B.

Communication and "A notification is displayed in response to V. INSIGHTS

notifications with determining that the one or more validation
messages rules exclude at least one constraint from the
A. Technology Alone is Not Omnipotent; Human-Machine
set of constraints." Collaboration Maximizes Strengths of Each Other’s
System software is not omnipotent. While it can operate
Visualizations with "Each block in the flowchart or block efficiently according to human-defined logic, it cannot
process flowcharts, diagrams may represent a module, segment, automatically adapt to business scenarios and changes that
figures, and diagrams or portion of instructions, which comprises
one or more executable instructions."
exceed its design requirements, still requiring continuous
iteration, optimization, and reconfiguration by humans.
Typical front-end pattern of interactive analysis in the
software include: Data mining is also not omnipotent. While it can uncover
hidden patterns from massive amounts of data, the quality,
1) Integrated Customized Detail Information Panel integrity, accuracy, consistency, and representativeness of the
This kind of panel is a customized display and operation data sources limit its predictive and decision-making accuracy,
panel for multi-source, multi-modal data. It provides business necessitating human iteration of algorithms, interpretation,
analysts with a front-end layout optimization and an and validation of results.
information presentation interface that aligns with human
Artificial intelligence is similarly limited. Though it can
reading habits. The panel can display images, graphs, text,
mimic certain human intelligent behaviors to a certain extent
interactive object entities, interactive diagrams, etc., and
with the support of big data and algorithms, it still relies on
includes operational buttons such as Verify, Project, Dismiss,
human intuition, wisdom, sensibility, and flexibility in areas
as shown in Fig.12-1.

774
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
such as cross-domain comprehension, creativity, moral aim to enhance the macro understanding, situational
judgment, and emotional communication. awareness, and change prediction capabilities of the dynamic
twin model, which is built upon a dynamic ontology and
Palantir has inherited the human-machine collaboration reflects the objective world, then it will lead to the formation
model from its founders' work on PayPal's security of an intelligent dynamic ontology.
authentication system [24], demonstrating the typical
application and validation of the human-machine symbiosis C. Innovating Interactive Visualization Analysis Tools
and collaboration concept in specific proprietary domains. Based on Big Data to Heuristically and Collaboratively
This article argues that in the business environment, Stimulate Collective Human Intelligence
clearly defining the functional boundaries and scopes of Palantir's developed and iteratively refined data analysis
human and system software, adopting human-centered design, products, widgets, and technologies (as shown in Fig. 5)
iteration, and optimization of technology, and achieving embody best practices in the field of data analysis, enabling
human-machine collaboration can maximize the strengths of better support for heuristic individual exploration and
both sides. On the one hand, it is necessary to establish human stimulating collaborative group analysis to a certain extent.
thinking patterns and interaction habits as the guiding This article believes that in the "human-machine"
principle, explore and research new interactive spaces formed information interaction, it is necessary to further study the
by Augmented Reality, Virtual Reality, brain-computer visualization of dynamic ontology models in Augmented
interfaces, etc. Technologies and tools that assist in enhancing Reality to assist human users in quickly and accurately
human users' insight generation and decision-making should forming decisions based on information and data in a way that
be studied, such as multi-modal information visualization and can be approximately encompassed by human vision and
interactive operation techniques associated with the ontology overarchingly understood by the human brain. Innovative
in AR space. On the other hand, within a unified data space research is needed in areas such as visual space information
based on dynamic ontology, it is essential to explore and visualization through Augmented Reality, high-sensitivity
research the application of AI and Agent technologies for virtual interaction operations in confined spaces, AI-assisted
automatic and autonomous subject relationship mining and multi-mode interaction decision-making technology. In terms
prediction, event expansion analysis and forecasting, etc. of improving the efficiency of "humans-machines", we should
B. Dynamic Ontology: Creating Digital Twins of Customer also conduct innovative research on visual interactive
Data and Models collaborative operation software and tools for a wider
geographical area, greater decentralization, more complex
Palantir adheres to the philosophical concept of ontology,
situations, and more teams and personnel, based on the
revolutionizing the traditional linear approach to data
analysis of business models and storage distributed big data
organization and processing. By constructing domain-specific
environments, such as distributed multi-seat human-machine
dynamic ontologies, it establishes mappings from data to
online analytical context sharing technology, temporary
ontologies and from ontologies to physical entities. These
offline environment analytical data caching and
mappings, validated and precise by humans, enable digital
synchronization technology, small group of adjacent units
twins across concepts, data, and processing logic. Instead of
temporary networking and interactive analysis technology,
working with fragmented, semantically impoverished, and
etc.
relationally unclear discrete data, customer business personnel
can now analyze and manipulate objects within dynamic VI. CONCLUSION
ontologies from their unique perspectives.
By analyzing the front-end analysis software within
IoT/
Real-time
Triditional Key Foundry Palantir's Gotham platform, this paper has identified the
Componets
primary capabilities and functionalities of the software. It has
ERP
preliminarily explored dynamic ontology technology, big data
CRM
system technical architectures, and human-centric interactive
visualization analysis techniques for big data. Furthermore,
insights have been gained regarding the benefits of human-
Foundry
computer collaboration and symbiosis in maximizing the
Digital Twin
Foundry Ontology strengths of both, the role of dynamic ontology in bridging
data and digital twins, and the innovation of human-centric
interactive visualization technologies. These findings lay a
solid foundation for subsequent in-depth research.

REFERENCES
Fig. 13. Presentation of Palantir Ontology Principles and Concepts [25] [1] Palantir. https://www.palantir.com/[EB/OL].(2023-06-14)
[2] Yibing Wang. The Rise of the Most Mysterious Unicorn: Palantir.
This article believes that ontology can reconstruct discrete https://new.qq.com/rain/a/20200705A0KZ1800[EB/OL].(2020-07-
data into information models that conform to human cognitive 06)
and thinking patterns, improving the efficiency of searching, [3] Uncovering the Core Technologies of Big Data Unicorn Palantir.
understanding, and generating insights from information and https://www.sohu.com/a/138760741_464065/[EB/OL].(2017-05-06)
data. Next, we need to conduct in-depth research on the [4] Dolphin Investment Research. Palantir ： "Mysterious" Military
technology of dynamic ontology construction for multi-source Weapons Activated by AI.
and multi-modal data. Furthermore, leveraging the advantages https://www.36kr.com/p/2448673066506372/ [EB/OL].(2023-09-26)
of current LLMs in intent understanding and reasoning, we [5] Palantir-Gotham. https://www.palantir.com/platforms/gotham/
[EB/OL].(2023-06-14)

775
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.
[6] Leqing Think Tank. The Mysterious Big Data Unicorn Palantir: The [16] Secure Collaboration.
Exclusive Visualization Service Provider for the CIA. https://www.palantir.com/offerings/defense/secure-
https://baijiahao.baidu.com/s?id=1594469698778410206&wfr=spider collaboration/features/ [EB/OL].(2024-10-08)
&for=pc [EB/OL].(2018-03-09) [17] Palantir-Foundry-DataIntegration.
[7] Is it Really a Good Business? Exclusive Decoding of Palantir's https://www.palantir.com/platforms/foundry/data-integration/
Differentiated Competitive Advantages | In-Depth Research Report [EB/OL].(2023-10-16)
(Part II). https://xueqiu.com/2445992253/256975914 [EB/OL].(2023- [18] Palantir-Attributes-Security.
07-31) https://www.palantir.com/attributes/security/ [EB/OL].(2023-10-22)
[8] Palantir-Foundry-Getting started. [19] Palntir-Palantir and Data Protection.
https://www.palantir.com/docs/foundry/getting-started/introductory- https://www.palantir.com/offerings/data-protection/#right-to-be-
concepts/ [EB/OL].(2023-08-26) forgotten-module [EB/OL].(2023-10-22)
[9] Na Wang, Zhihui Jiang. Overview of Domestic and International [20] Introduction to Palantir Gotham, the Big Data Analysis Platform of
Research Status on Dynamic Ontology Construction [J]. Journal of Palantir. http://www.d1net.com/bigdata/news/322330.html
Modern Information, 2020, 040(004):159-166. [EB/OL].(2014-12-15)
[10] Xiang Li，Kun Liu，Yi Liu，Rui Liu，Qiang Wang. Research on [21] Andrew Iliadis & Amelia Acker (2022). The seer and the seen:
Reconstruction Methods of Implicit Relationships in Operational Data Surveying Palantir's surveillance platform.
Oriented to Dynamic Ontology [C]. China Command and Control https://doi.org/10.1080/01972243.2022.2100851 [EB/OL].(2022-08-
Conference, 2024. 23)
[11] [Protégé. https://protege.stanford.edu/ [EB/OL].(2023-08-11) [22] Geocoding and geosearch. https://doc.arcgis.com/en/arcgis-
[12] [TopBraid Composer. https://topbraidcomposer.org/html/ [EB/OL]. online/reference/geocode.htm [EB/OL].(2022-09-06)
(2023-08-11) [23] What is geofence?. https://proximi.io/geofence-complete-guide-
[13] WebODE. https://github.com/oeg-upm/webODE [EB/OL].(2023-08- geofencing/ [EB/OL].(2022-09-06)
11) [24] Alpha Factory. Palantir: Cathie Wood's Favorite, Silicon Valley's Most
[14] Palantir-Foundry-The Platform-Ontology. Mysterious Big Data Analysis Company.
https://www.palantir.com/foundry-explained-get-demo/#ontology https://baijiahao.baidu.com/s?id=1700710271263079773&wfr=spider
[EB/OL].(2023-09-10) &for=pc [EB/OL].(2021-05-25)
[15] Registration Statement on Form S-1 of Palantir Technologies Inc.. [25] Palantir-Foundry-Power next-generation operations.
https://www.sec.gov/Archives/edgar/data/1321655/000119312520230 https://www.palantir.com/foundry-explained-get-demo/
013/d904406ds1.htm#rom904406_11 [EB/OL].(2023-10-08) [EB/OL].(2023-11-11)

776
Authorized licensed use limited to: Gangneung Wonju National University. Downloaded on March 13,2025 at 08:19:40 UTC from IEEE Xplore. Restrictions apply.