<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by David U. Ndukwu, Ph.D. on Medium]]></title>
        <description><![CDATA[Stories by David U. Ndukwu, Ph.D. on Medium]]></description>
        <link>https://medium.com/@undukwu?source=rss-2a977b6b20d3------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*yDuDrwTWrga6l62pbJX2Dg.jpeg</url>
            <title>Stories by David U. Ndukwu, Ph.D. on Medium</title>
            <link>https://medium.com/@undukwu?source=rss-2a977b6b20d3------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 08:18:21 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@undukwu/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Ontology-Driven Autonomous Data Pipeline Architecture]]></title>
            <link>https://medium.com/@undukwu/an-ontology-driven-autonomous-data-pipeline-architecture-d02f05c95d99?source=rss-2a977b6b20d3------2</link>
            <guid isPermaLink="false">https://medium.com/p/d02f05c95d99</guid>
            <category><![CDATA[data-architecture]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[agents]]></category>
            <dc:creator><![CDATA[David U. Ndukwu, Ph.D.]]></dc:creator>
            <pubDate>Mon, 02 Jun 2025 07:30:58 GMT</pubDate>
            <atom:updated>2026-07-30T06:59:01.427Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Introducing a revolutionary framework that transforms data engineering from manual orchestration to autonomous intelligence backed by Agentic AI workflows</em></p><h3>Introduction</h3><p>Imagine this scenario: It’s 2 AM, and your critical customer analytics pipeline has failed again! The data engineering team scrambles to diagnose cascading failures across seventeen interconnected ETL jobs. Meanwhile, the marketing team’s campaign launch is delayed by weeks because nobody understands the cascading effect of e.g. how the customer segmentation model connects to the product catalog schema and how the product catalog schema in turn connects to the ETL pipelines … Sounds familiar?</p><p>This scenario plays out thousands of times daily across enterprises worldwide. When ETL processes fail, the consequences ripple through IT departments and into the heart of business operations, leading to incorrect analytics, misguided strategic decisions, and costly operational disruptions. The primary objective of any ETL is to deliver business value, but this goal is often hindered by fragile systems. The fundamental problem isn’t technical complexity it is architectural philosophy.</p><p>Traditional data engineering treats pipelines as static plumbing: rigid, brittle, and requiring constant human intervention. But given the rise of Agentic AI, what if we could architect data pipelines differently i.e. by allowing them to understand their own purpose, adapt to changing requirements, and evolve autonomously? What if data pipelines could exhibit intelligence starting from requirements hardening to publishing production grade data assets for consumption?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SJpO566vWP2gf2Ri4Mu6_A.png" /></figure><p>Imagine a world where this approach results in efficiency gains that could be summarized in headlines such as below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gyaaJmOvHSPO7l6-uIHi-Q.png" /></figure><p>No doubt the gains are enormous and the possibilities are equally endless.</p><h3>The ontology foundation</h3><p>Modern data engineering operates on a fundamentally flawed assumption: that data movement and transformation can be separated from semantic understanding of business context. We build pipelines that move bits and bytes without comprehending meaning. This creates three critical failures:</p><ol><li><strong>Semantic Disconnection</strong>: ETL jobs transform data without understanding business context</li><li><strong>Unquantifiable Brittle Dependencies</strong>: Schema changes cascade unpredictably through downstream systems making it hard for impact to be quantified</li><li><strong>Knowledge Silos</strong>: Business logic remains trapped in individual engineers’ or analysts minds thus slowing down speed to market new business opportunities</li></ol><p>For example, consider the steps required to fulfil a request from the marketing team to add a new business metric e.g. “customer lifetime value by geographic region” — the data analyst completes the requirements mapping, the data engineer manually maps business concepts to technical schemas, writes transformation logic, and hardcodes business rules. When requirements change perhaps adding “seasonal purchasing patterns”, the entire pipeline would have to undergo a manual redesign and the process repeats all over again.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S70QpCtK4ZC2LDcQW2Tuew.png" /></figure><p>An ontology, in computational terms, is a formal specification of concepts and relationships within a domain (Gruber, T. R.; Guarino, N.). Unlike traditional schema definitions that describe structure, ontologies capture semantics — the <em>meaning</em> behind the data and the relationships or rules required to preserve such meaning . Ontologies help to answer questions such as:</p><ul><li>What is a “customer” and how does it relate to “purchase events”?</li><li>Which transformation logics preserve business meaning versus which violate it?</li><li>How do changes in one domain concept affect related concepts etc.?</li></ul><p>In this new paradigm, the ontology serves as the <strong>universal semantic foundation</strong> that every component references. Rather than hardcoded ETL logic, the proposition is to have autonomous agents that reason about data transformations using semantic understanding derived from datasets that surrounds the domain in view.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_oNvbo1sV178RJtLcohi1A.png" /></figure><h3>The ontology as a knowledge graph infrastructure</h3><p>The ontological foundation leverages knowledge graphs, specifically a <strong>Directed Cyclic Graph (DCG)</strong> akin to a hybrid architecture combining:</p><ol><li><strong>A Graph Database Layer</strong>: Stores the core ontological relationships, business rules, and semantic constraints.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_bdcRTBiC_CkE6-5unkrGQ.png" /></figure><p><strong>2. A Semantic Reasoning Engine</strong>: Validates transformations against ontological constraints and infers new relationships and publishes updates thereafter.</p><p><strong>3. A Version Control System</strong>: Manages ontological evolutions with branching, merging, and rollback capabilities, treating business knowledge like code end-to-end.</p><p>This infrastructure enables something unprecedented: <strong>data systems that understand their own purpose and can evolve accordingly when changes to business contexts are detected.</strong></p><h3>An Agentic AI framework: the six specialized intelligences</h3><p>Traditional data engineering relies on human orchestration of tool chains. Our framework replaces this with <strong>specialized AI agents</strong> that collaborate autonomously, each grounded in the shared ontological foundation.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BvhRXI4imI1T-SVCQ48Mmw.png" /></figure><p>The breakthrough insight is architectural: rather than linear ETL pipelines, we create an <strong>interconnected</strong> <strong>ecosystem</strong> where agents continuously learn from each other and update their knowledge base through the shared ontological foundation. This creates emergent intelligence that grows smarter over time and treats data pipelines as living components.</p><h4>Agent 1: The ontology ground truth (foundation agent) 🧠</h4><p><strong>Role</strong>: The cognitive core that <strong>continuously</strong> maintains and evolves semantic understanding of the data ecosystem.</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Knowledge Extraction</strong>: Automatically mines business rules from documentation, conversations, and usage patterns</li><li><strong>Ontological Evolution</strong>: Learns from system interactions to refine and expand semantic models with human approvals.</li><li><strong>Consistency Enforcement</strong>: Validates all other agents’ actions against semantic constraints</li></ul><p><strong>Technical Implementation</strong>: Combines large language models fine-tuned on domain ontologies with graph neural networks for relationship inference.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bunR9FMJcLXoZQUsuXopHQ.png" /></figure><h4>Agent 2: Requirements analysis agent 🏗️</h4><p><strong>Role</strong>: Translates business requirements into semantic specifications</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Natural Language Processing</strong>: Extracts semantic intent from user input business requests</li><li><strong>Impact Analysis</strong>: Predicts downstream effects of proposed changes using ontological reasoning created a priori</li><li><strong>Value Estimation</strong>: Quantifies business impact of data model modifications and cascaded downstream changes</li></ul><p><strong>Academic Foundation</strong>: Builds on recent advances in semantic parsing and neural-symbolic reasoning (Huang et al., 2023; Chen &amp; Manning, 2024).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/701/1*98tcOGWVZKsaVYbyFUfllw.png" /></figure><h4>Agent 3: Pipeline generation agent ⚙️</h4><p><strong>Role</strong>: Converts semantic specifications into executable data processing code for use in ETL pipelines</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Template-Based Generation</strong>: Creates pipeline code from reusable and validated ontologically-grounded patterns</li><li><strong>Performance Optimization</strong>: Automatically optimizes generated code based on data characteristics learned a priori</li><li><strong>Quality Integration</strong>: Embeds data quality checks derived from ontological constraints to preserve business meaning end-to-end</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/711/1*tpNEz5jrCob1IZMv2OhTog.png" /></figure><h4>Agent 4: Quality assurance agent 🔍</h4><p><strong>Role</strong>: Maintains data integrity through semantic validation</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Semantic Anomaly Detection</strong>: Identifies data quality issues by reasoning about ontological violations</li><li><strong>Automated Test Generation</strong>: Creates comprehensive test suites based on inferred semantic constraints</li><li><strong>Quality Contract Management</strong>: Establishes and enforces data quality contracts between data system components</li></ul><p><strong>Research Foundation</strong>: Extends recent work in probabilistic data quality assessment (Kumar et al., 2024, Ndukwu &amp; McIver, 2010) with ontological reasoning capabilities.</p><h4>Agent 5: Semantic layer propagation agent 📖</h4><p><strong>Role</strong>: Manages the translation between technical implementation and business understanding</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Documentation Generation</strong>: Automatically creates business-friendly documentation from technical specifications</li><li><strong>Change Impact Communication</strong>: Notifies stakeholders of changes using business terminology</li><li><strong>Semantic Consistency Validation</strong>: Ensures uniform understanding across all system components</li></ul><h4>Agent 6: Data assets publication and monitoring agent 🚀</h4><p><strong>Role</strong>: Publishes and manages intelligent data delivery to downstream analytics platforms</p><p><strong>Capabilities</strong>:</p><ul><li><strong>Usage-Informed Optimization</strong>: Adapts data distribution patterns based on consumption analytics</li><li><strong>Progressive Deployment</strong>: Implements safe rollout strategies with automatic rollback capabilities</li><li><strong>Performance Monitoring</strong>: Tracks end-to-end system performance and optimization opportunities</li></ul><h3>Summarising the novel contributions</h3><h4>1. A circular feedback architecture for data pipeline orchestration</h4><p>Traditional data pipelines are linear: extract → transform → load. The proposed framework introduces <strong>circular feedback loops</strong> where downstream consumption patterns <strong>incrementally</strong> inform upstream optimization viz:</p><ul><li>Quality anomalies that automatically refine ontological definitions</li><li>Usage patterns that optimize pipeline generation strategies</li><li>Business feedback that evolves to harden semantic understanding</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X5y7rEz1Y8rqTMtvEoeRCA.png" /></figure><p>Leveraging the autonomy of Agentic AI workflows, this creates <strong>self-improving systems</strong> that become more accurate and efficient over time.</p><h4>2. A confidence-based autonomy for risk mitigation</h4><p>Rather than a binary automation framework, the proposal seeks to conjecture a <strong>graduated autonomy</strong> based on semantic confidence scores:</p><ul><li><strong>High confidence (&gt;90%)</strong>: Fully autonomous execution</li><li><strong>Medium confidence (70–90%)</strong>: Automated with human notification</li><li><strong>Low confidence (&lt;70%)</strong>: Human-in-the-loop decision making</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JT1_zUho-90Fb8ocJed5Mw.png" /></figure><p>This balances innovation with risk management, enabling organizations to gradually increase system autonomy as trust on the AI Agents builds.</p><h4>3. Semantic versioning of business logic</h4><p>The framework extends software versioning concepts to <strong>business semantics </strong>as follows:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tSu63HB-QMvgFkQ2hCUm2A.png" /></figure><p>This kind of semantic versioning enables precise impact analysis and safe evolution of business logic — something that’s very hard to achieve with traditional approaches.</p><h4>4. A shift towards ontological test-driven development (OTDD)</h4><p>Traditional TDD focuses on functional correctness. The proposed framework introduces <strong>Ontological TDD (OTDD) </strong>where tests validate semantic consistency before data changes are committed by agents</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kk3n4H1nCcYxgwt52CHNrg.png" /></figure><h3>Conclusion</h3><p>For decades, we’ve treated data systems as sophisticated plumbing — moving and transforming information without understanding meaning. This approach has now reached its limits.</p><p>Today, we stand at an inflection point in data engineering due to the increasing adoption of Agentic AI workflows in automated task completion. This presents an opportunity to rethink traditional methods for data pipeline orchestration and create an ecosystem that’s self-evolving.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vwgZYGSiVj0VHxES3ysBDQ.png" /></figure><p>The implications extend far beyond efficiency gains, enabling data systems that understand their own purpose through:</p><ol><li><strong>Continuous learning: </strong>Adapting data pipelines dynamically as business contexts evolve to maintain alignment with desired outcomes</li><li><strong>Guided requirements elicitation: </strong>Anticipating business user needs while operating within defined ontological constraints</li><li><strong>Autonomous evolution: </strong>Self-improving through semantic reasoning that preserves core business meaning across iterations</li><li><strong>Self-healing operations: </strong>Proactively detecting and resolving issues through intelligent automation, minimizing downtime and reducing dependency on manual interventions</li><li><strong>Intelligent collaboration: </strong>Enabling seamless data team coordination through “Agentic AI Contracts” that define interaction boundaries and workflow rules</li></ol><p>This framework delivers substantial benefits by dramatically reducing the time advanced data teams need to generate high quality analytic insights. The underlying approach represents more than just the evolution of data engineering. It establishes the foundation for next-generation intelligent enterprise data ecosystems. The real transformation happens when thousands of data engineers, data architects, data analysts, technical leaders etc. embrace a semantic-first thinking and begin building systems that truly understand their purpose.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lgOw7xqmEAIQnMTIxO9vUQ.png" /></figure><h3>Disclaimer</h3><p><em>The concepts presented are my original ideas</em>, developed with assistance from <strong>Claude AI (Sonnet 4)</strong> for reasoning validation and presentation refinement.</p><h3>References</h3><ul><li>Chen, D., &amp; Manning, C. (2024). “Neural-Symbolic Reasoning for Natural Language Understanding.” <em>Journal of AI Research</em>, 78, 245–267.</li><li>Gruber, T. R. (1993). “A Translation Approach to Portable Ontology Specifications.” <em>Knowledge Acquisition</em>, 5(2), 199–220.</li><li>Guarino, N. (1998). “Formal Ontology and Information Systems.” <em>Proceedings of FOIS’98</em>, 3–15.</li><li>Huang, L., et al. (2023). “Semantic Parsing with Large Language Models: A Comprehensive Study.” <em>ACL 2023 Proceedings</em>, 1234–1245.</li><li>Kumar, S., et al. (2024). “Probabilistic Data Quality Assessment in Large-Scale Systems.” <em>VLDB Proceedings</em>, 47, 567–580.</li><li>Ndukwu, U., &amp; McIver, A. (2010). “An expectation transformer approach to predicate abstraction and data independence for probabilistic programs.” <em>EPTCS Logic in Computer Science</em>, 28, 129–143.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d02f05c95d99" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Scaling Data Products Delivery Using Domain-Oriented Data Pipelines]]></title>
            <link>https://medium.com/data-science/scaling-data-products-delivery-using-domain-oriented-data-pipelines-869ca9461892?source=rss-2a977b6b20d3------2</link>
            <guid isPermaLink="false">https://medium.com/p/869ca9461892</guid>
            <category><![CDATA[devsecops]]></category>
            <category><![CDATA[notes-from-industry]]></category>
            <category><![CDATA[editors-pick]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[data-pipeline]]></category>
            <dc:creator><![CDATA[David U. Ndukwu, Ph.D.]]></dc:creator>
            <pubDate>Mon, 27 Sep 2021 13:31:21 GMT</pubDate>
            <atom:updated>2022-09-01T02:03:50.487Z</atom:updated>
            <content:encoded><![CDATA[<h4><a href="https://towardsdatascience.com/tagged/notes-from-industry">Notes from Industry</a></h4><h4>A tested approach for the rapid delivery of data products at scale.</h4><h3>Introduction</h3><p>The recent reimagining of “data as a product” i.e. a slight deviation from the well-known mantra of “data as a strategic asset”, has necessitated the revamp of modern data pipeline architectures to support the rapid delivery of data solutions at scale. This proposition is correct irrespective of the underlying data architecture upon which the target enterprise is defined.</p><p>The concept of “data as a product” is increasingly gaining momentum amongst organizations that have huge amounts of data at their disposal. This would typically include giant retailers, social media platforms and financial services firms, just to mention a few. While most of these organizations desire a transition to this new paradigm or ways of working, unfortunately, most of them can’t, simply because of the limitation posed by their existing data pipeline architectures i.e. they are monolithic in nature.</p><p>Scaling the delivery of data products in modern data ecosystems therefore necessitates a complete overhaul of existing monolithic data pipelines. This should be preceded by a thorough understanding of the various interactions (aka interface contracts) that exist between the domains that contribute to the evolution of the data product. Empowering these domains with dedicated pipelines is key for a successful data product delivery. Further reading about the construct of “data domain ownership” as well as related principles governing data as a product can be found in Dehghani’s recent publication <a href="https://lnkd.in/g6jAFgx9">here</a>.</p><h3>What is a domain-oriented data pipeline?</h3><p>As stated above, the philosophy underpinning data as a product enforces domain ownership on the components of the data products. This also includes the data pipelines that orchestrate the build, test and deploy of the domain components. It is important to mention that while these pipelines may be independent in their own rights, they are however not completely isolated. As such, <em>a data pipeline architecture that enables the respective domains in a data product team to independently build, test and deploy re-usable components of the data product using a minumum set of shared baseline artefacts is known as a domain-oriented data pipeline</em>. It is important to mention that this definition assumes an architecture that implicitly incorporates best practices for modern data pipeline designs e.g. security, scalability, obeservability etc.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KVB1qwexWnb7u-a-A5lnTQ.png" /><figcaption>High-level architecture of a domain-oriented data pipeline</figcaption></figure><p>The high-level architecture depicted above captures the essential components of a domain-oriented data pipeline. The key principles outlined below are very important to help maximize the benefits delivered by a domain-oriented data pipeline:</p><ol><li><strong>Create the right team topology: </strong>As seen in the diagram above, the pipeline design is predicated on the <strong>domain teams</strong> responsible for building out various components of the data product (e.g. Domain Team 1 … Domain Team N). Also, since re-use is a key feature of a domain-oriented data pipeline, the team topology should support a <strong>core product team</strong> responsible for embedding common standards that can be leveraged by the domain teams as they build out their respective data workloads. For example, the core product team is responsible for baking and publishing known versions of base images of the data product for consumption by the domains in the ecosystem.</li><li><strong>Minimize duplication and maximize re-use via shared baseline components: </strong>This comes on the back of the understanding of the relationships between the various domain components. A domain-oriented data pipeline architecture should encourage re-use and minimize duplication across the pipeline ecosystem as much as possible<strong>.</strong> While this can be a balancing act, the overall benefits far outweigh the initial efforts with the domain components relationship mapping exercise.</li><li><strong>Define clear domain regions (including asset nomenclatures): </strong>While the intention is to ensure that each domain pipeline exists independently (even though not completely isolated), proper attention has to be given to how separation of concerns is achieved for the domains wrt the shared components. This is done by creating regions of interactions for each domain. Furthermore, the regions must be identifiable by clear nomenclatures (i.e. naming standards) especially on shared components such as storage, pipeline build artefacts, code repositories etc. This clarity will come in handy for ease of data products evolution and consumption by end-users.</li><li><strong>Define clear domain boundaries (including ownerships and accountabilities): </strong>A lack of boundaries, ownerships and accountabilities is always a recipe for disaster in any pipeline architecture. This problem can be further exacerbated in a domain-oriented pipeline architecture. However, if clearly defined, a domain-oriented data pipeline architecture has the benefit to deliver a <strong>federated governance model</strong> on the data ecosystem. That is, the domains as well as the core product teams will have the flexibility to define structures that create common standards and help to eliminate potential friction in the overall team topology.</li></ol><h3>Benefits of domain-oriented data pipelines</h3><p>No doubt scaling data products with traditional monolithic data pipelines can be very challenging. Domain-oriented data pipelines on the other hand, tend to simplify the delivery approach by adding the following benefits (there are many more):</p><ol><li>They are very easy to scale both from a technology and business standpoint. Imagine the convenience of adding a new pipeline to support a brand new business domain from a pre-packaged domain pipeline template.</li><li>The independence of the domain pipelines largely helps to eliminate a single point of failure in the entire data product delivery process.</li><li>Domain-oriented data pipelines make it easier to detect performance bottlenecks in the data product delivery lifecycle since metrics can be measured and collected at the level of each domain.</li><li>They deliver a federated data governance model which gives control to the actors in the team topology to own and be accountable for their respective pipeline components.</li></ol><h3>Conclusion</h3><p>A data product, if developed correctly can be elegant at first sight. However, a “true data product” is not just defined by its elegance but also on its ease of maintenance as well as its ability to scale in order to support the changing needs of the enterprise. This is where domain-oriented data pipelines come in.</p><p>The principles described herein stem from my recent experience leading a team of data pipeline engineers to architect and implement a domain-oriented data pipeline for an enterprise. So while some of them may not be directly applicable to your data ecosystem, the bulk of it however holds true for most modern data ecosystems though with a few tweaks. For example, even though only one <strong>core product team</strong> has been depicted in the reference architecture, in reality, this team can be scaled out with dedicated pipelines, depending on the complexity of the data product.</p><p>Finally, it is important to mention that data, at the right grain, must be made available to the actors in a domain-oriented pipeline architecture in order to support build and test of the various data components prior to having them published to the shared zones for consumption in the ecosystem.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=869ca9461892" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/scaling-data-products-delivery-using-domain-oriented-data-pipelines-869ca9461892">Scaling Data Products Delivery Using Domain-Oriented Data Pipelines</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>