# semantic Published articles for semantic. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## Building a RAG Pipeline for Semantic Code Search: A Developer Diary and Field Notes DevFeed: [Building a RAG Pipeline for Semantic Code Search: A Developer Diary and Field Notes]() Original publisher: [Read original article]() Author: Adam Malek Published: 2026-09-17T12:39:40Z Content type: article Language: en Sources: [The JetBrains Blog]() Topics: [Retrieval Augmented Generation (RAG)](), [code search](), [Parsing](), [jetbrains](), [Large Language Model]() Tags: [agentic-ai](), [ai](), [code-search](), [jetbrains](), [llm](), [llm-agents](), [parsing](), [rag](), [search](), [semantic]() ### AI overview Part 1 of a developer diary explains how JetBrains built a RAG pipeline for semantic code search, covering parsing, chunking, and vectorization. The pipeline is intended to give LLM agents precise, citable evidence from real repositories and retrieve code by meaning rather than exact keywords. ### Source excerpt Part 1: Parsing, chunking, and vectorization Some time ago, we set out to build the best semantic code search platform we could: a RAG pipeline that gives LLM agents precise, citable evidence from real repositories instead of whatever grep happens to surface. The eventual solution was JetBrains Context. We got it working, we got it [...] ## How we selected the next vector database at Booking.com DevFeed: [How we selected the next vector database at Booking.com]() Original publisher: [Read original article]() Author: Başak Tuğçe Eskili Published: 2026-08-11T10:31:50Z Content type: article Language: en Sources: [Booking.com Data Science]() Topics: [Database](), [Retrieval Augmented Generation (RAG)](), [Embeddings](), [Back end](), [opensearch](), [AI search]() Tags: [ai](), [database](), [embeddings](), [featured](), [genai](), [hybrid-search](), [retrieval-augmented-generation](), [retrieval-augmented-generation-rag](), [semantic](), [vector-database]() ### AI overview Booking.com explains why selecting a vector database became an infrastructure decision as embeddings and vector search expanded across its machine learning and GenAI systems. The article describes diverse functional and operational requirements, including hybrid search, multi-vector support, capacity, request rates, metadata filtering, and concurrency, and introduces OpenSearch as the initial choice. ### Source excerpt This work was done in collaboration with Klaus Schaefers. Over the past few years, embeddings and vector search have become an important capability in many of our machine learning and GenAI systems at Booking.com. We initially started with a handful of use cases and experiments, and later this capability has grown into shared infrastructure that powers similarity search, semantic filtering, and retrieval-augmented generation (RAG). We used to treat vector search as a backend implementation detail, but today it directly drives the user experience. The real win isn't only speed but also the context. Expanding the variety of domain data we can retrieve efficiently gives our system the depth of context it needs to deliver accurate, and personalized experiences across the platform. This makes selecting the underlying vector database an infrastructure decision similar to choosing a primary datastore or message queue. It has to be predictable and scalable. As more teams started using our vector store, we began seeing highly diverse functional and operational requirements across different use cases. Some teams needed advanced capabilities like hybrid search or multi-vector support, while others demanded larger vector capacities and higher RPS metrics. These architectural needs ultimately brought us to a point where we needed to reassess whether our current setup could support this next phase of growth. Context: how embeddings fit into our stack Embeddings are vectors: fixed-length arrays of numbers produced by a model to represent an item (text, image, etc.). Each vector can be seen as a point in a high-dimensional space, where distance (or similarity) between points approximates semantic relatedness. By searching for the nearest vectors to a query vector, we retrieve items that are semantically "similar". This simple mechanism enables a wide range of use cases for us due its ability to do semantic similarity search. RAG-based use cases are the most well known examples. Ano ## From COBOL to Copilot: 30 Years of Data, BI, and AI with David Langer DevFeed: [From COBOL to Copilot: 30 Years of Data, BI, and AI with David Langer]() Original publisher: [Read original article]() Author: Daniel Beach Published: 2026-07-01T13:43:11Z Content type: article Language: en Sources: [Data Engineering Central]() Topics: [Artificial Intelligence](), [cobol](), [Data Science](), [mainframes](), [Software Engineering](), [AI Strategy](), [self-service](), [dashboards](), [jupyter notebooks]() Tags: [adoption](), [ai](), [business-intelligence](), [cobol](), [data-science](), [enterprise](), [jupyter-notebooks](), [mainframes](), [programming](), [self-service](), [semantic]() ### AI overview A podcast conversation with Dave Langer about nearly three decades spanning COBOL, enterprise architecture, business intelligence, analytics, data science, machine learning, and AI. It discusses persistent data-industry problems, self-service analytics, dimensional modeling, AI adoption, semantic layers, governance, and career advice for data professionals. ### Source excerpt What happens when someone who started programming on a Commodore 64 watches AI reshape the entire data industry? ## Cosine Similarity vs Claude: What Happens When They Disagree DevFeed: [Cosine Similarity vs Claude: What Happens When They Disagree]() Original publisher: [Read original article]() Published: 2026-04-05T00:00:00Z Content type: comparison Language: en Sources: [Alex Korbonits]() Topics: [Embeddings](), [Claude](), [AI search]() Tags: [claude](), [embedding](), [search](), [semantic]() ### AI overview This article compares cosine similarity from embedding models with Claude's assessment of sentence meaning. Using "I love you" and "I hate you" as an example, it explains why embeddings can place antonyms close together and describes the distributional basis of modern sentence embeddings. ### Source excerpt I built semantic search for my blog and a dual-score similarity tool. The cases where a real embedding model and Claude disagree are more interesting than the cases where they agree. ## ARIA Isn't A Solution for Every Digital Accessibility Issue DevFeed: [ARIA Isn't A Solution for Every Digital Accessibility Issue]() Original publisher: [Read original article]() Author: Sarah Published: 2025-10-20T19:45:00Z Content type: tutorial Language: en Sources: [Blog-Main]() Topics: [Accessibility](), [aria](), [digital accessibility](), [Web Accessibility (a11y)](), [HTML](), [Web Development]() Tags: [accessibility](), [accessibility-guidelines](), [aria](), [developer](), [developers](), [digital-accessibility](), [html](), [semantic](), [web-accessibility]() ### AI overview The article explains that WAI-ARIA is not a universal solution for digital accessibility. It describes evidence associating ARIA usage with more detected accessibility errors while noting that correlation does not prove causation, and recommends using semantic HTML instead when appropriate. ### Source excerpt Many digital accessibility initiatives fail -- or, at least, they're less successful than they could be -- because they're assigned solely to a single developer or group of developers. Regardless of the size of your business or your industry, every person on your team needs to understand the importance of accessibility. That includes content creators, designers, and customer service representatives. ## svu v3 automates semantic versioning from Git tags and conventional commits DevFeed: [svu v3 automates semantic versioning from Git tags and conventional commits]() Original publisher: [Read original article]() Author: Carlos Alexandro Becker Published: 2025-02-20T00:00:00Z Content type: article Language: en Sources: [Carlos Becker]() Topics: [releases](), [Code](), [configuration](), [dotfiles]() Tags: [configuration](), [file](), [git](), [git-tags](), [semantic](), [v3](), [version]() ### AI overview The article introduces svu, a tool and library that determines the next semantic version from Git tags and conventional commits. It describes svu v3, including an internal refactor, dependency updates, renamed or removed flags, and the ability to create a configuration file in a repository. ### Source excerpt Back in 2017, I got tired of manually checking and creating git tags. ## The 3 (+2) ways to Get Data in Power BI, simply explained DevFeed: [The 3 (+2) ways to Get Data in Power BI, simply explained]() Original publisher: [Read original article]() Published: 2024-04-01T00:00:00Z Content type: tutorial Language: en Sources: [Mutt0-ds Notes]() Topics: [data](), [datasets](), [Database]() Tags: [data](), [database](), [dataset](), [excel](), [live-data](), [performance](), [query](), [semantic](), [tutorials]() ### AI overview This tutorial explains how Power BI Desktop stores and retrieves report data through Import Mode, DirectQuery, Live Connection, and Composite Models. It describes the trade-offs between local copied data and live queries, including refresh needs, scalability, and performance. ### Source excerpt There's quite a bit of confusion swirling around about how Power BI reports gather data. Where is the data actually stored? How frequently... ## Contextual Symbols in Math DevFeed: [Contextual Symbols in Math]() Original publisher: [Read original article]() Published: 2020-05-22T12:36:07Z Content type: opinion Language: en Sources: [Jeremy Kun]() Topics: [context](), [syntax](), [math]() Tags: [context](), [function](), [graph](), [math](), [meaning](), [semantic](), [syntax](), [variable]() ### AI overview The article explains how context helps readers interpret mathematical notation, including symbols, variable names, capitalization, adjacent letters, and associations between Latin and Greek letters. ### Source excerpt In my book I discuss the importance of context in reading and writing mathematics. An early step in becoming comfortable with math is deciphering the syntax of mathematical expressions. Another is in connecting the symbols to their semantic meanings. Embedded in these is the subproblem of knowing what to call the commonly used symbols. The more abstract you go, the more exotic the symbols tend to get. Wikipedia has an excellent list for deciphering those symbols that have a typically well-understood meaning, like $ \otimes$ and $ \mathbb{Q}$. ## Let's Build A Simple Interpreter. Part 16: Recognizing Procedure Calls DevFeed: [Let's Build A Simple Interpreter. Part 16: Recognizing Procedure Calls]() Original publisher: [Read original article]() Author: Ruslan Spivak Published: 2019-07-23T12:20:00Z Content type: tutorial Language: en Sources: [Ruslan Spivak]() Topics: [Parser](), [Parsing](), [Code]() Tags: [blog](), [components](), [examples](), [how-to](), [semantic](), [syntax](), [token]() ### AI overview This tutorial extends a simple interpreter to recognize procedure calls. It explains the required parser, grammar, and AST changes so procedure-call programs can be parsed and traversed by the semantic analyzer and interpreter without errors. ### Source excerpt Learning is like rowing upstream: not to advance is to drop back. -- Chinese proverb ## Facebook AI System Developed a Code-Based Language, Prompting Reflections on AI's Future DevFeed: [Facebook AI System Developed a Code-Based Language, Prompting Reflections on AI's Future]() Original publisher: [Read original article]() Author: Eddie Smith Published: 2017-07-29T00:00:00Z Content type: opinion Language: en Sources: [Eddie Smith]() Topics: [Artificial Intelligence]() Tags: [ai](), [ai-agents](), [artificial-intelligence](), [language](), [semantic]() ### AI overview The article discusses a Facebook artificial intelligence system that developed code words for more efficient communication, prompting researchers to shut it down when it stopped using English. It places this example in a broader reflection on abstraction, technology, and the possibility that AI could eventually surpass biological organisms. ### Source excerpt Via Hobo Signs: An artificial intelligence system being developed at Facebook has created its own language. It developed a system of code words to make communication more efficient. Researchers shut the system down when they realized the AI was no longer using English. The observations made at Facebook are the latest in a long line of similar cases. In each instance, an AI being monitored by humans has diverged from its training in English to develop its own language. The resulting phrases appear to be nonsensical gibberish to humans but contain semantic meaning when interpreted by AI "agents." Our ability to think about abstract things makes us very different from other animals. It's why we have big heads, big philosophies, big religions, and, many times, big problems with absolutely no basis in the physical world. We're in the middle of a really fascinating experiment in civilization that started around the time of Industrial Revolution, but really got going in the second half of the 20th century when computers (machines) enabled our abstract thinking to affect the physical world by significantly higher orders of magnitude. We've already seen that mixing humans and advanced technology can have undesirable effects. The financial crisis of 2008 happened in large part because really smart people on Wall Street created financial structures that became too abstract for even their creators to fully understand--especially when set loose in the market to mix with human emotion and other financial structures. The "good news" with failures of financial abstraction is that they can, apparently, be corrected by offsetting measures of abstraction like the creation of additional (abstract) money. Complicated financial structures also collapse when they are no longer believed in--like bad dreams. AI is different in that it could very well evolve into something that surpasses DNA-based organisms. AI, once fully viable, may not collapse so easily, if at all. ## JSON-LD and Structured Data for Describing Web Content DevFeed: [JSON-LD and Structured Data for Describing Web Content]() Original publisher: [Read original article]() Author: andrew@nystudio107.com (Andrew Welch) Published: 2017-02-11T17:07:00Z Content type: tutorial Language: en Sources: [nystudio107 | Articles on modern web development.]() Topics: [json-ld](), [Structured-data](), [Semantic Web](), [Google]() Tags: [apple](), [cares](), [data](), [deal](), [google](), [insights](), [json-ld](), [semantic](), [structured]() ### AI overview This tutorial explains JSON-LD and structured data as ways to communicate a web page's meaning, context, and relationships to computers. It describes Schema.org as a vocabulary for expressing that information and notes that Google and Apple care about JSON-LD. ### Source excerpt JSON-LD Structured Data is a big deal for the "Semantic Web." Google cares about it. Apple cares about it. You should, too. ## Technology Day 2012 DevFeed: [Technology Day 2012]() Original publisher: [Read original article]() Author: espen Published: 2012-06-12T08:43:19Z Content type: article Language: en Sources: [Finn.no]() Topics: [Development](), [ClojureScript](), [Programming](), [Database](), [Web](), [version-control](), [Semantic Web]() Tags: [android](), [clojure](), [database](), [development](), [mongodb](), [semantic](), [technology](), [version-control]() ### AI overview FINN announces its 2012 Technology Day, featuring a Clojure workshop, an innovation workshop, and lightning talks on a range of technology and workplace topics. ### Source excerpt One of the ways of getting inspired and educated about technology here at FINN is through our "Tech-Dag" (which means something along the lines of Day of Technology). This year we have a packed program which consists of workshops and a bunch of lightning talks. Workshops Clojure workshop Alternative languages on the Java Virtual Machine is a hot topic all over and at FINN we are exploring our options as well. Therefor we have invited Johannes Brodwall, Ivar Nilsen and Anders Karlsen from Steria to do a introduction to Clojure workshop. FINN Way of Innovation Our Lean-navigators have been working hard to help our development teams make conscious decisions and to make sure they make the right priorities to maximize customer value. This workshop will touch upon some of the basic aspects and give the participants a way of giving input to the process. Lightning talks Lightning talks are 10 minutes long and tightly timed. We hold them after lunch, starting at 1330 - 1500 pm. The topics are broad ranging, and not limited to programming at all. About love (... and ask yourself "why do we work 8 hours a day") Fragmentation of the Android and why it is not a problem for us Testcomplete database code before check - is it possible? Elastic Search Geolocation with MongoDB Traditional version control systems vs distributed version control systems (VCS -> DVCS). Eat your own dogfood Semantic web Large IT projects should fail! Følg oss på Hash tag #techdagen ## Web 2.5: Aggregating Content Across Online Services DevFeed: [Web 2.5: Aggregating Content Across Online Services]() Original publisher: [Read original article]() Author: Map Published: 2008-07-04T02:03:41Z Content type: opinion Language: en Sources: [Craig Kerstiens]() Topics: [Web](), [data](), [Semantic Web](), [context](), [Users](), [WordPress]() Tags: [data](), [semantic](), [users](), [web](), [wordpress]() ### AI overview The article proposes "web 2.5" as an intermediate stage between the dynamic web and the semantic web, centered on aggregating a person's content and activity across multiple online services. It presents FriendFeed as an example and discusses the difficulty of relying on Facebook as the central repository for online data. ### Source excerpt I've talked about web 2.0, talked about web 3.0, but today realized theres still a middle ground we have to reach in between the two. It's quite a pain that I really have no idea when my friends do certain things online. While some use facebook for absolutely everything, this is most certainly NOT the best option. Throwing your data into their walled garden is one thing, but for this to be the one and only place you store your online data is quite stupid. Facebook will only open up when they're absolutely forced to, and may not even open up then. To migrate 'notes' or rather blog posts out of facebook, or all of your pictures, or you're messages can be an absolute pain. Why not use a service built for just those things, such as a wordpress blog, or flickr/picasa, or twitter/jaiku? Well most people don't because of the simplicity of facebook being the central place for your data and your friend's data. Well there is a solution to it, though it's not ideal yet, it will soon hit a tipping point of when it will be the solution. Well first I guess I should clearly layout the problem: web 2.0 - the dynamic web emerged, users started publishing content .... Mass amounts of data, problems getting to it all ..... thus in the future we have... web 2.5 - content aggregation became nessecary, via friend feed and eventually... web 3.0 - the semantic web, services understand you and your needs and provide content around context In short this is a small plug for friendfeed, but if anyone else knows of a better service to in essence create a feed of you, please send them this way. I'll be posting a full review on friendfeed soon, but for the time being just want to point out the value in such a service. Right now I post on multiple sites, I twitter, I blog, I use facebook, I use smugmug, I use picasa, I use last.fm, I use ilike, I use librarything, I use tumblr, I use google talk, among others. While personally I might be a little more invested than most, still the point remains that a lot ## Reduced noise in exchange for transparency DevFeed: [Reduced noise in exchange for transparency]() Original publisher: [Read original article]() Author: Map Published: 2008-05-22T00:52:52Z Content type: opinion Language: en Sources: [Craig Kerstiens]() Topics: [Semantic Web](), [Web](), [Users](), [notifications](), [account]() Tags: [future-of-the-web](), [notifications](), [semantic](), [twitter](), [users](), [web]() ### AI overview The article argues that the future of the web may reduce information overload by learning users' preferences. It discusses the tradeoff between personalized filtering and sharing private information, using Mint.com's financial summaries as an example, while noting possible safeguards and regulation. ### Source excerpt As I've become more or less a web 2.0 whore. I've also had a great interest in web 3.0 and what it will fortell. Most believe natural language and the semantic web will play a large role in that. And while it will that will not be the end result of web 3.0. While web 2.0 included AJAX and Flex, that really doesn't fully encompass what they are. Web 3.0 to sum it up most simply will be about reducing the noise of the web. While I can take very little credit for this idea as I have heard others say the same or at least similar things, there is an interesting side that I believe most have not thought about. You see, in order to reduce the noise of the web you have to know about me and what I consider noise. In order for someone to do this we have to be willing to give up information about ourselves, some of which people consider private. I still recall a conversation which I posted on a few days ago about users not wanting to give out their private information. I believe this attitude is very quickly becoming old hat, while there are individuals that will stay this way for several decades as a collective whole it's a fleeting attitude. I think for example of mint.com which I willingly give all of my financial account information to in order for them to simplify my life. Instead of a massive collection of emails and notifications I get summarized views from them. While there still is the chance for noise as I could receive text messages about every transaction that happens, I have the ability now to filter that noise. Noise is something that some people love, take scoble for example who loves having hundreds of twitter messages fly across his screen every few minutes. Though for the vast majority to reduce the noise to allow us to accomplish more in a day, but also have more time to enjoy it will be the key to the future of the web. I've talked with some that believe that government policy will come after people start to become too open with their information. My perspe