HPE (Hewlett Packard Enterprise) has announced a new offering and marketplace infrastructure with the enviable title of HPE Haven OnDemand Combinations, designed to accelerate the development of “next generation cognitive applications.” The combinations are preconfigured sets or chains of APIs which have been available from HPE in the IDOL on-premise software and more recently offered in a variety of freemium and commercial packages through HPE’s Haven OnDemand cloud services.
HPE’s description of the new functionality can be found here:
Any application developer who needs to access and process rich and/or disparate data sets has long been familiar with the need to use a pipeline processing approach, in which source data needs to be accessed, then transformed for analysis or integration with other streams of data, then subjected to a variety of analytics routines, each of which seek to enrich the aggregate dataset to render specific decision-making or business monitoring insights.
An example might be the challenges involved in developing an application to access professional and user-generated video news content, where the developer needs to access a variety of rich media formats, use voice recognition or closed caption text to render the audio into text streams, time-stamp the text to link the audio and video streams, use classifiers to group content items into categories by subject matter, use entity extractors to identify people, places, and things mentioned in the content, perhaps use sentiment analytics to interpret positive or negative sentiment expressed in the text data, or even use image analysis to gauge sentiment based on facial expressions in the video streams – to name just a few of the operations the developer might need.
The value proposition HPE is targeting with Combinations is to make the process of calling and sequencing API toolsets faster and easier by chaining APIs commonly used together and publishing them as preconfigured sets. Developers using the 70+ IDOL APIs will now be able to integrate clusters of them into custom applications more efficiently and with higher confidence that a specific Combination will work as expected.
At the same time, HPE is promoting a Combination Marketplace, with an initial catalog populated by HPE, where any developers can publish custom Combinations for their own private re-use or for public consumption across the community of Haven developers.
HPE’s IDOL assets represent the legacy of some of the earliest unstructured data analysis systems to incorporate the machine learning and neural network approaches which have become core building blocks in many analytics systems. It remains for HPE to leverage once again the value of this functionality and its institutional knowledge around the demands of cognitive applications.Share