<?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 Greg West on Medium]]></title>
        <description><![CDATA[Stories by Greg West on Medium]]></description>
        <link>https://medium.com/@greg-west?source=rss-dde1db540e17------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*hkpiuLCnCxAFoYJ2cp0sDA.jpeg</url>
            <title>Stories by Greg West on Medium</title>
            <link>https://medium.com/@greg-west?source=rss-dde1db540e17------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:21:36 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@greg-west/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[Why Knowledge Graphs Are Central to Industry 4.0 Initiatives]]></title>
            <link>https://greg-west.medium.com/why-knowledge-graphs-are-central-to-industry-4-0-initiatives-4e81a1560617?source=rss-dde1db540e17------2</link>
            <guid isPermaLink="false">https://medium.com/p/4e81a1560617</guid>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[industry-4-0]]></category>
            <category><![CDATA[semantics]]></category>
            <category><![CDATA[knowledge-graph-embedding]]></category>
            <category><![CDATA[knowledge-graph]]></category>
            <dc:creator><![CDATA[Greg West]]></dc:creator>
            <pubDate>Thu, 23 Feb 2023 20:58:17 GMT</pubDate>
            <atom:updated>2023-02-23T20:58:17.211Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UoboDQqYF1gXiO8L" /><figcaption>Photo by <a href="https://unsplash.com/@klalit?utm_source=medium&amp;utm_medium=referral">Lalit Kumar</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>We are in the midst of a major technological inflection point: knowledge graphs have developed into a foundational component of the modern Industry 4.0 technology stack. I have this perspective because in my role at Cambridge Semantics I constantly engage with business and technology leaders and discuss data strategies, initiatives, and future-forward planning. I’m not alone in this viewpoint; the software industry sees this market opportunity too, and most major technology vendors have <a href="https://www.assemblymag.com/articles/97286-digital-manufacturing-platforms-in-industry-40">positioned offerings.</a></p><p>Knowledge graphs are not a “new” technology. For years, proponents of knowledge graphs have touted their data integration capabilities. Today, knowledge graphs are increasing automation and reducing decision cycles, drastically optimizing data usage and efficiently driving down costs. This is because knowledge graphs semantically harmonize access to diverse data sources, resulting in machine readable and human understandable data. Now, the suite of demands and challenges in Industry 4.0 environments have pushed beyond traditional data landscape capabilities and propelled market interest toward knowledge graphs.</p><p>Before diving into the current explosion of interest, let me quickly establish the key requirements for the knowledge graph stack.</p><p>Key Requirements:</p><ol><li>The ability to construct an ontology or semantic model which describes the entities and relationships existing in the system.</li><li>The ability to integrate data from diverse sources into an RDF-based knowledge graph in alignment with the ontology.</li><li>The ability to execute performant queries on that data. This requirement is often overlooked, but Architectures which don’t include a distributed database with an OLAP (online analytical processing) design are inherently limited.</li></ol><p>Set those aside for a minute, and let’s discuss some key factors driving the growth of knowledge graph technologies.</p><p><strong>Manufacturing Systems and Processes Have Complex Relationships</strong></p><p>Digital twins are inherently intricate, combining data from several sources as they reflect an entity’s state across its lifecycle. The state of the entity and its relationships cannot always be easily modeled using a traditional schema. For example, a bill of materials for a given design has a hierarchical relationship between entities and sub-assemblies within a particular entity which present a challenge for a relational database. A knowledge graph is a natural way to express these relationships and create validation rules to ensure compliance.</p><p>Another complex use case is the enterprise supply chain which can be modeled in a network (graph) form. Supply chains are akin to Complex Adaptive Systems which feature ongoing emergent and divergent relationships. The data architecture supporting the robust supply chain must be adaptable and gracefully cope with change as a constant. Knowledge Graphs — especially those built on W3C OWL/RDF — are ideal for such systems.</p><p><strong>Semantic Reasoning Augments Traditional AI</strong></p><p>Knowledge graphs have long been understood to be particularly useful in analyzing specific types of systems. Graph algorithms provide a natural means of calculating shortest paths, system throughput, community detection, node importance, and node similarity (graph embeddings).</p><blockquote>The Graph Search Neural Network [is] a way of efficiently incorporating large knowledge graphs into a vision classification pipeline. We show in a number of experiments that our method outperforms standard neural network baselines for multi-label classification. -<a href="https://www.ml.cmu.edu/research/phd-dissertation-pdfs/kdmarino_phd_machinelearning_2021.pdf">Dr. Kenneth Marino, PhD Thesis</a></blockquote><p>What’s become increasingly clear is that these techniques can also augment traditional machine learning. Researchers and academics have realized that robotic vision algorithms can benefit from logic available solely through semantic reasoning. One early adopter, Amazon, is betting that this technology can eventually lead to a <a href="https://www.therobotreport.com/amazon-testing-robots-to-transport-oversized-items-in-fulfillment-centers/">fleet of autonomous robots.</a> Having a semantic understanding of the fulfillment center is optimizing the movement of robots as they navigate the busy floor and in the future may help with robotic grasping and other tasks.</p><p><strong>Knowledge Graphs Enable Q &amp; A Engagement</strong></p><p>Artificial Intelligence is certainly having a moment. ChatGPT has captured the public’s attention, garnering headline articles in the New York Times and other newspapers. Personally, I’m more interested in the capabilities of the GPT-3 model (third-generation Generative Pre-trained Transformer). Over the past few months, I’ve experimented with GPT-3’s capabilities via their public API and I was stunned. I provided as input a semantically described ontology; from this GPT-3 was able to generate SPARQL queries from natural language questions. I tested with non trivial questions that require multiple-hop relationships amongst connected entities as well as aggregations. Because the ontology is represented with semantic terms, GPT-3 was able to translate synonyms or derivatives of the language I used onto the semantic model. An example query (input) and the automatically generated output is below.</p><pre>Input: What are the components that occur more than twice in the bill of <br>materials for the tubes with attached sensors that have a temperature reading<br>over 140?<br><br>SELECT ?component<br>WHERE {<br>  ?tube stm:p_Has_Sensor ?sensor .<br>  ?sensor stm:p_Has_Reading ?reading .<br>  ?reading stm:p_Temperature ?temp .<br>  ?bill stm:p_For_Tube ?tube .<br>  ?bill stm:p_Has_Components ?component .<br>  FILTER (?temp &gt; 140)<br>}<br>GROUP BY ?component<br>HAVING (COUNT(?component) &gt; 2)</pre><p><a href="https://jens-lehmann.org/files/2022/ieee_access_sgpt.pdf">Researchers from all over the world are trying the same thing.</a> The implications of this are manifold. The next logical step is the introduction of a simple chatbot or user interface where managers or technicians can ask any question of a manufacturing system and receive real-time data in response.</p><p><strong>A Vision For the Future</strong></p><p>At this point, many companies are still cobbling together their knowledge graph strategy, but there are some firms in the vanguard.</p><p>Bosch is one company that has gone all-in on <a href="https://www.bosch-connected-industry.com/de/en/portfolio/bosch-semantic-stack">knowledge graph and semantic technology</a>. Not surprisingly, Bosch is one of Cambridge Semantics largest clients. Anzo, our knowledge graph product, is the only full stack knowledge graph platform with a massively parallel processing (MPP) architecture that allows users to create knowledge graphs by overlaying and combining data from any data source — structured or unstructured. Each data source is activated in the RDF graph engine as an in-memory data layer, on top of which additional layers can be added that logically connect, extend, and transform the knowledge graph.</p><p>Without exception, these three use cases can benefit significantly from this architecture:</p><p><em>Complex Manufacturing Systems</em></p><p>Creating an ontological model of a complex system isn’t a straightforward exercise, and mapping data onto that model can be even trickier. The Anzo Graph Data Interface (GDI) can do a parallel load directly from enterprise data sources to RDF, and then data layers can perform in-memory transformations. This means that regardless of the shape, complexity, or dirtiness of source data, it can be loaded sight unseen, then restructured and cleansed using powerful transformation queries. This stands in stark contrast to the approach of most label property graph (LPG) databases which will require pre-defining the schema and creating pipelines that extract and transform source content in a specific format.</p><p><em>Semantic Reasoning</em></p><p>Semantic reasoning itself is a computationally expensive exercise, and generally requires several iterations to perfect. Anzo’s data layers streamline the exercise of testing different permutations of modeling and connecting the knowledge graph, which can accelerate the generating of one of the most important inputs to AI. Furthermore, Anzo offers a set of out of the box algorithms and a framework where data scientists can implement user defined extensions (UDXs) that leverage the MPP performance of the underlying graph engine.</p><p><em>Q&amp;A Engagement</em></p><p>The nature of questions enterprise leaders generally ask are often complex, as in computationally and logistically expensive. In the example query I posted above, all sensor values for all the tubes in the system need to be scanned, and then the common components determined. The execution of that query benefits significantly from Anzo’s distributed OLAP architecture. Unlike data virtualization approaches that require managing network load and caching, Anzo creates a single in-memory database layer. Application end users can receive punctual answers and downstream operational systems aren’t impacted by the stress of complex questions.</p><p>If you have thoughts on additional use cases or would just like to have an interesting discussion on the possibilities of knowledge graph please message me at <a href="/service/mailto:greg@cambridgesemantics.com">greg@cambridgesemantics.com</a> or visit our website: <a href="http://www.cambridgesemantics.com/">www.cambridgesemantics.com</a> .</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e81a1560617" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Applying Graph Algorithms on Knowledge Graphs:]]></title>
            <link>https://greg-west.medium.com/applying-graph-algorithms-on-knowledge-graphs-1d94a070fb98?source=rss-dde1db540e17------2</link>
            <guid isPermaLink="false">https://medium.com/p/1d94a070fb98</guid>
            <category><![CDATA[graph-algorithms]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[data-fabric]]></category>
            <category><![CDATA[knowledge-graph]]></category>
            <category><![CDATA[graph-database]]></category>
            <dc:creator><![CDATA[Greg West]]></dc:creator>
            <pubDate>Tue, 19 Jan 2021 14:43:24 GMT</pubDate>
            <atom:updated>2021-01-20T16:11:53.289Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>Finding Influencers in a Real-World Social Network</em></h4><p><em>As an engineer at </em><a href="https://www.cambridgesemantics.com/"><em>Cambridge Semantics</em></a><em> over the past 7 years, I’ve had numerous opportunities with my colleagues to implement graph analytics on an enterprise scale knowledge graph.</em></p><h4>Introduction</h4><p>Growing up outside of Boston, I was always fascinated by the rich history of the area. One of the most famous examples was the ride of Paul Revere, who at the onset of the American Revolution warned colonists of the impending British Army’s march to seize the militia’s gunpowder and arms in outlying towns. His success carrying the message was due to the fact that he was one of the most socially connected people in Boston.</p><p>Finding the most influential members in a network continues to be an important need — and there are a variety of graph algorithms that can identify them. The classical method is to define a graph structure, essentially a Labeled Property Graph (LPG) schema. Data is conformed to the LPG model prior to algorithm execution. At Cambridge Semantics, with our product Anzo, we take a more novel approach that is detailed below. Anzo creates an in-memory Resource Description Framework (RDF) knowledge graph based on the original data sources. Users can then create in-memory “derived graphs” to implement algorithms.</p><p><strong>Overview — A Real World Social Network</strong></p><p>Consider a RDF knowledge graph based on the Linked Data Benchmark Council’s (LDBC) benchmark which describes a social network.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*pHiygIrdqoLCaX2keatW8g.png" /><figcaption>Fig 1: Social network in a real-world model.</figcaption></figure><p>Note that it does not indicate the strength of relationships between individuals or even which people are friends. That is because RDF-based knowledge graphs represent knowledge by linking information originating from siloed data sources. Crucially, they represent data as it actually exists, not in an idealized structure, custom-built for a specific analysis. This is a key difference from LPG databases, where experts define a specific graph structure and create custom rules to shape incoming data to that structure.</p><p>RDF graphs can also use graph algorithms to determine who the most influential people are from this model, with a couple simple preliminary steps. In this example, we use Betweenness Centrality to detect key influencers in a social network.</p><h4>Betweenness Centrality — What is it?</h4><p>Betweenness Centrality measures nodes in a network that hold the most information transmission power. Most social networks will break down into clusters, with key leaders at the focal points. A small subset of individuals will be members of several clusters. These individuals are said to have high betweenness centrality, as they can quickly transmit key information to leaders of several groups. Paul Revere had a high degree of betweenness centrality in colonial Boston, as he had personal connections to leaders of major revolutionary groups.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*Um-uhzF0sxht8AJAY4iMwA.png" /><figcaption>Fig 2: Example network graph. The red nodes have the highest betweenness centrality.</figcaption></figure><h4>The Real World Mechanics</h4><p>To run Betweenness Centrality, a couple of preparatory steps are necessary in which we generate a derived graph from our real-world graph model. First, we need to extend the model to establish links between people that the algorithm will use. From an ontological perspective, we need to add a new predicate to the model. We can even leverage RDF* to give this connection a weight that indicates the strength of that relationship.</p><pre>&lt;&lt;&lt;person1&gt; :isConnectedTo &lt;person2&gt;&gt;&gt; :weight &lt;value&gt;.</pre><p>Next, we need to establish a heuristic that instantiates this relationship between individuals in our graph. Studying the model, logically one pattern indicating such a relationship is one individual posting a message, and another individual replying to it. We can use the number of messages (posts and comments/replies) between two individuals to compute a weight value to add to this edge. The more messages they exchange, the higher the strength of this connection. As depicted in the figure below, we apply a SPARQL query to the original knowledge graph, and the inserted statements produce a derived graph specifically designed to answer our original question using the chosen algorithm. All that’s left is to actually execute the algorithm, by far the easiest part.</p><pre>INSERT {<br>    &lt;&lt;?person1 :isConnectedTo ?person2&gt;&gt; :weight ?numMessages.<br>}<br>WHERE {<br>    {<br>        SELECT ?person ?person2 (COUNT(?message) AS ?numMessages)<br>        WHERE {<br>            ?message :hasCreator ?person1.<br>            ?comment :replyOf ?message.<br>            ?message :hasCreator ?person2.<br>        }<br>        GROUP BY ?person ?person2<br>    }<br>}</pre><h4>How does this scale?</h4><p>The benefit of our approach is only realized if it can be applied to very large knowledge graphs comprising many sources. Fortunately, Anzo’s massively parallel processing (MPP) architecture allows users to combine data coming from numerous siloed sources to build and process knowledge graphs at unparalleled scale. A key feature of Anzo is that each data source is activated in the RDF graph engine as an in-memory data layer, on top of which additional layers can be added that logically connect, extend, and transform the knowledge graph. These additional layers are often implemented as a set SPARQL queries that describe the instructions that perform in-memory transformations and create, as in our example, derived graphs.</p><p>There are two major benefits that our method using data layers enables for executing graph algorithms. First, in the real world, users will want to rapidly test and apply different combinations of logic to create derived graphs and then execute algorithms. For example, perhaps the query above could be modified to take advantage of additional information the knowledge graph contains such as where people work or went to school. Data layers make the process of testing the heuristics and algorithms iterative. To test a change, all that is necessary is an in-memory refresh of the layer which can occur in moments. Second, the ontology and data structures do not need to be modified for each use case. If, in addition to trying to find influencers, the desired output was clusters of individuals based on forum activity, this activity could be prototyped in a new layer independent of the results of the first. This novel approach allows users to rapidly develop new use cases once the baseline knowledge graph has been established.</p><p>I’ve described one way that a specific graph algorithm can be applied in a sample RDF knowledge graph. Our team at Cambridge Semantics has decades of experience developing and applying knowledge graphs in real-world scenarios. If your interest is piqued and you’d like to have a conversation, please message me at <a href="/service/mailto:greg@cambridgesemantics.com">greg@cambridgesemantics.com</a> or visit our website: <a href="http://www.cambridgesemantics.com/">www.cambridgesemantics.com</a>.</p><h4>Sources:</h4><ol><li>The original inspiration for the intro is Malcolm Gladwell’s book <em>Tipping Point</em>. For a little more background, see this blog: <a href="https://medium.com/better-marketing/paul-revere-a-brilliant-genius-of-viral-marketing-8ac62edc459d"><em>https://medium.com/better-marketing/paul-revere-a-brilliant-genius-of-viral-marketing-8ac62edc459d</em></a></li><li>If you are interested on the technical details for different centrality algorithms, see this slide deck: <a href="https://ocw.mit.edu/courses/civil-and-environmental-engineering/1-022-introduction-to-network-models-fall-2018/lecture-notes/MIT1_022F18_lec4.pdf">https://ocw.mit.edu/courses/civil-and-environmental-engineering/1-022-introduction-to-network-models-fall-2018/lecture-notes/MIT1_022F18_lec4.pdf</a></li><li>Kieran Healy has this very technical blog discussing how centrality algorithms would have identified Paul Revere as a person of interest in colonial Boston: <a href="https://kieranhealy.org/blog/archives/2013/06/09/using-metadata-to-find-paul-revere/">https://kieranhealy.org/blog/archives/2013/06/09/using-metadata-to-find-paul-revere/</a></li><li>Figure two is an excellent visual of betweenness centrality. Again, from a highly technical source: <a href="https://cs.hse.ru/data/2015/05/14/1098547089/4._Centrality_Metrics.pdf">https://cs.hse.ru/data/2015/05/14/1098547089/4._Centrality_Metrics.pdf</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1d94a070fb98" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>