Showing posts with label Google cloud. Show all posts
Showing posts with label Google cloud. Show all posts

Tuesday, 21 November 2017

Google Cloud Natural Language API

Cloud Natural Language API
Google’s Cloud Natural Language API

Cloud Natural Language API, established by Google is said to provide customers with language analyser which according to the company `reveals the structure as well as the meaning of your text. The public beta launch of Cloud natural language API is a new service giving developers access to Google-powered emotion analysis, entity, recognition together with grammar analysis.

 Some of this tends to gauge believing some words positive and the others negative. When observed by Motherboard it was found that the analyser of Google interpreted some words like homosexual to be negative. This is evident that the API that tends to judge depending on the information it has been fed, now seems to give out partial analysis. The tool has been developed to provide companies with a preview on how their language would be expected.

Editing complete sentences would provide predictive analysis on each word and as the overall statement on a negative to positive scale, respectively. AI systems have been trained in utilisation of texts, media as well as books given to it.

Whatever Cloud Natural Language API consumed to form its criteria in assessing English text for sentiment, it influenced the study to negative attribution of certain descriptive terms. No confirmation has been provided by Google to Motherboard as to the body of the text fed to the Cloud Natural Language API.
 
API Connects Other Pre-Trained Machine Learning API
 
Once it begins to engage content from the outside world even if begins with a remote set of contents to comprehend sentiments, it tends to get polluted with the negative word connections found in it. A confirmation had been given by Google to Motherboard that its NLP API had produced biased results in a statement.

There had been clear parallels with the ill-fated as well as impressionable AI chatbots Tay of Microsoft, which had been rapidly, pulled offline by the company in March 2016 after the users of Twitters had taught it to be a shockingly racist as well as sexist conspiracy philosopher.

In July, Google had tried once again with its bot Zo which had learned the same horrible habits form human and had to be quickly shut down. The new API connects the other pre-trained machine-learning API of Google such as the Cloud Speech API which has been made available in public beta, together with the Vision API and the Translate API.
 
Assist Text in English/Spanish/Japanese
 
The latest Cloud Natural Language API presently tends to assist texts in English, Spanish and Japanese. The purpose of Google here is to provide a service which could meet the scale as well as the performance essential for developers and enterprises in a comprehensive range of industries. Providing API for sentiment analysis and entity recognition is not new where services like Thomson Reuters Open Calais have been providing assistance for entity recognition for around ten years now. Sentiment analysis is also not a new concept. On the other hand, grammar analysis API which tend to classify parts of speech and develop dependency analyse trees are not as extensively available still. It would be interesting to know how developers would be utilising these apps though it is easy to see how the same could be utilised to power chat bots for instance and support them in comprehending incoming request.

Friday, 21 July 2017

Google Antes Up Its Own Cloud Migration Appliance

Google Cloud

Google is bringing its own data transfer appliance for cloud migration


When we talk about the cloud migration hen the toughest challenge to overcome is to ensure reliable and consistent migration methodology. Moving databases and data centers are not an easy game even for the proficient administrators. This is faced by almost all the major companies and new start-ups when they are trying to build new application or make use of new data residing in the cloud. Data migration between two points is always seen as tough egg to break due higher costs and huge time consumption this is where cloud vendors come into the play.

Remedy for data to cloud migration problem


The problem faced in the cloud migration is quite incomprehensible even with modern technologies at disposable. If a person has 10 Gbps connection then transferring petabyte of data from any data center will consume as many as 12 days to put it on the cloud. In the old golden days companies used to Sneakernet in the sky wherein a pile of data is loaded on a secure disk and it is shipped off to any of their cloud vendor. Microsoft Azure made use of this system for quite some time before a new system came into being called Snowball and this method is also used by the Amazon AWS. This doesn’t mean that sending secure disks to the cloud vendor has become obsolete.

Google is moving into the enterprise cloud migration business


The tech giant is getting into the enterprise cloud business on a serious note but quite specifically in the migration appliance space. In this end of the business the consumer arc is very similar to online shopping portal. User goes online, orders the device and this particular device is made available for definite set of time after which they have to send it back to the provider.

Cloud migration market is buzzing with immense order in the one petabyte category therefore Google is introducing two models with size of 100 and 480TBytes. On other hand its competitors Amazon tends to work in lower end of the spectrum with 3 models with data size of 50, 80 and 100 TBytes units. Amazon is way ahead in the field of data transfer and it has developed a number of solutions for high end data migration with Amazon Snowmobile wherein it brings a 12 wheeler 45-foot container for transferring 100 PBytes of data.

Pricing will be a key to grab a piece of the cloud migration market


Google is mainly focusing on cornering market for the petabyte cloud migration. Therefore it has brought on par pricing with Snowball for the smaller 100 TByte units and in order to make migration appliance more appealing to the users by keeping it 35% below the rival offerings. When it comes to design Google is going for plug-in based form and function while Amazon brings self-standing units.

However Google has revealed many details of its devices, service offerings, capabilities and benefits but it seems like it is eagerly looking forward to give stiff competition to market leaders namely Amazon and Microsoft Azure.