I'd like to say that I was highly organized, that I knew where every box ended up and what was in each box. Most people who move know the feeling of living in boxes even after the movers have left, the days spent dodging labyrinths of teetering cardboard, their arms and legs scored with paper cuts where they misjudged that one particular stack that seemed to have taken on a malicious life of its own. In retrospect, I've decided I'm going to go truly high-tech next time: buy a batch of beacons, one for each box, my laptop open as I carefully pack each cardboard container with my sundry possessions, adding each item into a personal knowledge graph so that I can tell exactly where everything in my new house, organized by topic, by room, by owner. I will gleefully take screenshots showing how masterful my graph-fu skills are for future articles, and maybe, just maybe, I wouldn't then have to sleep on the couch at night because I inadvertently packed the family cat. Ah, who am I kidding?

I help lead a working group focused on personal knowledge graphs (PKGs). Lately, it's functioned as a discussion and demo evaluation group for new technologies and how they might be used in a knowledge graph context. Different individuals want to annotate different kinds of data. Some do a lot of research. For them, the need is to annotate the links and associated text (in a simple and ideally machine assisted way from research sources so that machines can help retrieve the right links later on and discover (or rediscover) related links.

Hypernymy plays a fundamental role in many AI tasks like taxonomy learning, ontology learning, etc. This has motivated the development of many automatic identification methods for extracting this relation, most of which rely on word distribution. We present a novel model HyperBox to learn box embeddings for hypernym discovery. Given an input term, HyperBox retrieves its suitable hypernym from a target corpus. For this task, we use the dataset published for SemEval 2018 Shared Task on Hypernym Discovery. We compare the performance of our model on two specific domains of knowledge: medical and music. Experimentally, we show that our model outperforms existing methods on the majority of the evaluation metrics. Moreover, our model generalize well over unseen hypernymy pairs using only a small set of training data.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. The Justice Department charged a California man of hurling threats of violence against the Massachusetts-based Merriam-Webster Inc., while he allegedly accused the dictionary of promoting "lies and anti-science propaganda" regarding its gender definition entries for the words "woman" and "girl." Jeremy David Hanson, 34, of Rossmoor, California, was charged in federal court in Springfield, Massachusetts, by criminal complaint with one count of interstate communication of threats to commit violence. He was arrested and made an initial federal court appearance in the Central District of California on Wednesday.

Biomedical knowledge graphs (BioMedKGs) are essential infrastructures for biomedical and healthcare big data and artificial intelligence (AI), facilitating natural language processing, model development, and data exchange. For many decades, these knowledge graphs have been built via expert curation, which can no longer catch up with the speed of today's AI development, and a transition to algorithmically generated BioMedKGs is necessary. In this work, we introduce the Biomedical Informatics Ontology System (BIOS), the first large scale publicly available BioMedKG that is fully generated by machine learning algorithms. BIOS currently contains 4.1 million concepts, 7.4 million terms in two languages, and 7.3 million relation triplets. We introduce the methodology for developing BIOS, which covers curation of raw biomedical terms, computationally identifying synonymous terms and aggregating them to create concept nodes, semantic type classification of the concepts, relation identification, and biomedical machine translation.

Augmenting pre-trained language models with knowledge graphs (KGs) has achieved success on various commonsense reasoning tasks. However, for a given task instance, the KG, or certain parts of the KG, may not be useful. Although KG-augmented models often use attention to focus on specific KG components, the KG is still always used, and the attention mechanism is never explicitly taught which KG components should be used. Meanwhile, saliency methods can measure how much a KG feature (e.g., graph, node, path) influences the model to make the correct prediction, thus explaining which KG features are useful. This paper explores how saliency explanations can be used to improve KG-augmented models' performance. First, we propose to create coarse (Is the KG useful?) and fine (Which nodes/paths in the KG are useful?) saliency explanations. Second, to motivate saliency-based supervision, we analyze oracle KG-augmented models which directly use saliency explanations as extra inputs for guiding their attention. Third, we propose SalKG, a framework for KG-augmented models to learn from coarse and/or fine saliency explanations. Given saliency explanations created from a task's training set, SalKG jointly trains the model to predict the explanations, then solve the task by attending to KG features highlighted by the predicted explanations. On three commonsense QA benchmarks (CSQA, OBQA, CODAH) and a range of KG-augmented models, we show that SalKG can yield considerable performance gains -- up to 2.76% absolute improvement on CSQA.

While the natural language processing (NLP) field has been growing at an exponential rate for the last two years -- thanks to the development of transfer-based models -- their applications have been limited in scope for the job search field. LinkedIn, the leading company in job search and recruitment, is a good example. While I hold a Ph.D. in Material Science and a Master in Physics, I am receiving job recommendations such as Technical Program Manager at MongoDB and a Go Developer position at Toptal which are both web developing companies that are not relevant to my background. This feeling of irrelevancy is shared by many users and is a cause of big frustration. In general, however, traditional job search engines are based on simple keyword and/or semantic similarities that are usually not well suited to providing good job recommendations since they don't take into account the interlinks between entities.

It's been ten years since Google (now a child of holding company Alphabet) coined the term "knowledge graph" and described (in general terms) how their knowledge graph worked. And it's been over 20 years since Tim Berners-Lee, James Hendler and Ora Lassila published their first article to describe the semantic web they envisioned. Many knowledge graphs have been built using the semantic standards the W3C subsequently put in motion a decade or more ago. It's interesting to ponder what's happened since. Over the past decade, Alphabet has grown consistently to become one of the top six companies globally to achieve a market capitalization (total stock value of shares outstanding) of over $1 trillion.

Every day, millions of articles and papers are published. While there is a lot of knowledge hidden in those articles, it is virtually impossible to read all of them. Even if you only focus on a specific domain, it is still hard to find all relevant articles and read them to get valuable insights. However, there are tools that could help you avoid manual labor and extract those insights automatically. I am, of course, talking about various NLP tools and services. In this blog post, I will present a solution of how you can combine the power of NLP with knowledge graphs to extract valuable insights from relevant articles automatically.

I love constructing knowledge graphs from various sources. I've wanted to create a government knowledge graph for some time now but was struggling to find any data that is easily accessible and doesn't require me to spend weeks developing a data pipeline. At first, I thought I would have to use OCR and NLP techniques to extract valuable information from public records, but luckily I stumbled upon UK Gazette. The UK Gazette is a website that holds the United Kingdom's official public record information. All the content on the website and via its APIs is available under the Open Government License v3.0.