Category: Computing & data (Page 62 of 80)

Computing and data is a broad category. Our coverage of computing is largely limited to software, and we are mostly focused on unstructured data, semi-structured data, or mixed data that includes structured data.

Topics include computing platforms, analytics, data science, data modeling, database technologies, machine learning / AI, Internet of Things (IoT), blockchain, augmented reality, bots, programming languages, natural language processing applications such as machine translation, and knowledge graphs.

Gilbane Advisor 7-14-20 — perceiving, DSM, web 3.0, microservices

July 14, 2020 / Frank Gilbane

Dear Reader:

I hope all is well.

We have been busy updating our website and I thought you deserved a quick update. In mid-May we woke up “NewsShark” and re-activated our curated news service which hasn’t been active for a while. It is available on our site here, as a feed, and on Twitter. We publish news multiple times a week, and will check with you at some point to see if you are interested in an email version. We have consolidated all of our content on our main site, improved site navigation, added back search, and have a new simplified category structure – all available from any page. Finally, we are using schema.org markup and experimenting with some additional features that it allows — you’ll notice some of them as you poke around. We’ll update you as we formally roll them out.

Now to this issue’s recommended reading…

Comparing human and machine perception

This article is a wonderfully clear and concrete example of how easy it is to incorrectly interpret data from comparisons between deep neural networks and human perceptions, and how to think about further experiments to expose potential misinterpretations. There is also a broader lesson here for evaluating machine learning algorithms.

There is a link to the full paper, but this summary by the authors is a valuable resource for non-specialists. Read More

Decentralized web developer report 2020

The decentralized web is an amorphous collection of technologies and projects that are not a near-term threat to today’s imperfect and increasingly centralized web. But it is encouraging to see so much activity dedicated to a more open web, and this report by Fluence Labs’ Evgeny Ponomarev is an excellent way to get a feel for the landscape of the players, the challenges, and what software engineers, researchers, and others think. This is not one of those promotional market research reports, and doesn’t gloss over the challenges. The raw survey data is included. Read More

The seven deceptions of microservices

Software architectures are not the sort of thing you create or change lightly. Even if you’re convinced a different approach would be better, there are inevitably unforeseen developmental and operational consequences / costs which can quickly multiply scarily as a function of the number of moving parts. Software architects and experienced software engineers know this, but the whole team should understand the pros and cons of such a change. Software engineer Scott Rogowski suggests some things to watch out for when considering moving to a microservices development model. Read More

Online content sharing – pay to play?

Article 17 of Directive (EU) 2019/790 on Copyright in the Digital Single Market (the “DSM Directive”), introduces a new content management and liability regime for online content-sharing service providers (“OCSSPs”) … Article 17 is one of the most controversial provisions of the DSM Directive. Its supporters view Article 17 as facilitating more licensing of copyright protected works online to generate remuneration for rightholders whose works are shared by users on profit generating online platforms, while its detractors argue that it goes too far and will have an adverse effect on freedom of expression and the proper functioning of copyright exceptions online. Read More

Also…

Not that any of you need this, but maybe someone you know… CMU researchers develop an automatic politeness engine for text-based communications via Techcrunch
Caught in the mushy middle: How Quartz fell to earth via digiday
Do you really know what “RESTful” is supposed to mean? Roy Fielding’s Misappropriated REST Dissertation via TwoBitHistory.org
From the level-setting AI department… Where are all the robots?… via Techcrunch

The Gilbane Advisor curates content for content technology, computing, and digital experience professionals. We focus on strategic technologies. We publish more or less twice a month except for August and December. We do not sell or share personal data.

Subscribe | Feed | View online | Privacy policy | Editorial policy

Exasol launches Exasol V7

July 8, 2020 / NewsShark

Exasol launched Exasol V7 giving users the ability to run analytical models on larger data volumes to find more business-critical answers. Exasol V7 bolsters AI/ML model training with GPUs, improve the performance of their Data Vault models, and improves the use of unstructured data.

GPUs – Supporting the use of GPUs, Exasol V7 provides the speed and performance needed when training and retraining AI/ML models on large data sets and remove barriers to entry for deep learning.
Data Vault 2.0 – In Data Vault 2.0, all keys are stored as hashes. These hashes help quickly join and compare data from multiple tables and schemas, to improve query performance, create better user experiences and open up previously prohibitive analysis.
Semi-structured Data (JSON Function) – Exasol V7 natively supports multiple data formats (structured and semi-structured) within one database engine. In Exasol V7, JSON functions are natively integrated in SQL and can be executed directly in the database without the need for User-Defined Functions (UDFs).

Exasol V7 will be available on July 28, 2020.

https://www.exasol.com/en/v7/

DefinedCrowd launches DefinedData, a marketplace of AI datasets

July 7, 2020 / NewsShark

DefinedCrowd announced the launch of DefinedData, an offering that enables customers to accelerate their AI-initiatives by acquiring pre-collected, annotated, and validated AI training data from an online catalog. Customers can browse pre-collected AI datasets in multiple languages, domains, and recording types and either request samples or purchase. Customers can also choose between a one-time purchase or annual subscription that provides access to all of the new datasets. The catalog is expected to grow to include over 25,000 hours of speech and natural language data. Multiple key performance indicators (KPIs) will be used including Word Error Rate, gender distribution levels, age distribution, ambient noise levels, nativeness (accuracy of native speakers), and domain accuracy.

https://www.definedcrowd.com/

TransPerfect to share multilingual data set with AI and chatbot developers

July 6, 2020 / NewsShark

TransPerfect, provider of language and technology solutions for global business, announced that it is sharing a multilingual data set that will help AI and chatbot developers provide information on COVID-19 in multiple languages. The data set will allow researchers and technology organizations to train smart chatbots that could increase the coverage and appropriateness of responses provided by systems integrated into websites, social media services, virtual assistants, and other platforms. The training data set, which is available in English, Spanish, Arabic, and Mandarin Chinese, accessible at no charge. TransPerfect DataForce specializes in building AI experiences through data collection and annotation. The group has a broad pool of global resources and advanced technologies that gives it the ability to aggregate and package data that can be highly useful in improving responsiveness to COVID-19.

https://www.transperfect.com

Nuxeo unveils low-code user interface to open AI to business users

July 2, 2020 / NewsShark

Nuxeo announced a new, low-code user interface (UI) for Nuxeo Insight, its artificial intelligence (AI) service that enables companies to use their own data and content to train custom machine learning (ML) models. The new interface allows business users to easily create, train, deploy and administer personalised machine learning models. The interface offers a guided point-and-click experience, and users can also deploy and administer these models in production use cases. Nuxeo’s new user experience walks users through the step-by-step process of training and deploying customer machine learning models:

Select the various values and data types they want the new model to populate
Apply different parameters for predictions
Select the appropriate content and data for successful model training
Perform any necessary renditions or transformations

Nuxeo Insight’s new UI includes a dashboard that provides real-time updates on the training process and outcomes. It also enables organisations to administer their custom ML models, allowing them to quickly promote new models into production and to actively monitor the performance of their models over time. With the new UI, users can compare the performance of different ML models or different versions of the same model. The overall goal is to provide an intuitive, easy-to-use environment that enables Nuxeo customers to continuously improve the performance of their ML models.

https://www.nuxeo.com

pCloud now lets users decide where their files are stored

July 1, 2020 / NewsShark

pCloud is giving users the opportunity to select location of their data. This will give users greater control over the security of their files. Once the choice of where to store the data is made during registration – in the US or Europe – it is practically impossible to transfer them without the user’s knowledge or permission. Currently, the option to select the server location is available only to newly registered users. The company is developing the option to change the storage location for existing users and it will be available very soon.

The company has recently added a data centre in Luxembourg, meeting the needs of consumers and businesses even in the most regulated sectors in terms of consumer data protection and information security. pCloud’s new European data center is certified to SSAE 16 SOC 2, Type II standard. In addition, pCloud meets standards such as ISO 9001: 2015 (Quality Management Systems) and ISO 27001: 2013 (Information Security Management Systems). When using the pCloud service, user data is transferred via TLS / SSL protocols, which provide data protection during transmission.

https://www.pcloud.com/data-regions/

Informatica launches collaborative technology initiative with The ADAPT Centre

July 1, 2020 / NewsShark

Informatica announced a collaborative technology initiative with The ADAPT Centre, a Science Foundation Ireland (SFI) Research Centre for Digital Media Technology. The ADAPT Centre, funded by SFI, focuses on developing next-generation digital technologies that transform how people communicate by helping to analyze, personalize and deliver digital data more effectively. Informatica’s R&D team collaborate with ADAPT to leverage its applied research in Artificial Intelligence (AI) and Natural Language Processing (NLP) to accelerate the application of AI and Machine Learning (ML) to automate Informatica’s enterprise cloud data management offerings. Organisations are increasingly leveraging extremely large data sets to inform business decisions but have a range of obligations under regulations like The General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA), making data management even more complex. Informatica’s engineers have already developed many AI and ML capabilities as part of the Informatica CLAIRE engine to help customers use data effectively for their analytics, data governance and privacy, and customer experience (CX) needs.

https://www.informatica.com

Obie launches Browser extension to accelerate sharing of workplace documents

June 30, 2020 / NewsShark

Obie announced the launch of its new browser extension to democratize access to Obie’s core search, access, and knowledge sharing functionality. They also launched Personal Pro and Personal Free plans that decouple Obie’s search functionality from Slack for the first time to expand availability for individuals and remote teams. Obie uses natural language processing (NLP) to understand complex queries, as well as machine learning (ML) to improve results with every document search. Users can also manually add “FAQs” to store templates, text snippets, and frequently accessed information by simply highlighting the information in their browser and adding it to Obie.

https://obie.ai