The New Microservice in Lucidworks Fusion 5.0 Enables Easier Operationalizing of Models via Containers and APIs
Lucidworks, a leader in AI-powered search, announced native Python support in Fusion 5.0, the latest version of the company’s flagship enterprise search product. The new feature opens up the option for data scientists to train and build models using their preferred machine learning and deep learning libraries while at the same time streamlining the way they can handoff the models for production into the Fusion index and query pipelines.
Organizations are increasingly relying on data scientists and machine learning. However, the process of driving value from algorithms has been human-intensive. In many instances, once developed and trained, algorithms need to be translated/recoded into production system languages (like Java) before they can be deployed for indexing and querying.
Now, data scientists can use their favorite tools, like Scikit learn, SpaCy and Tensorflow to develop models for a wide spectrum of NLP, Machine Learning and Deep Learning techniques and immediately deploy them through a native Python connector into Fusion. By shortening the time to production and integrating deeper the data science tasks into search processes, organizations will drive more speed, scale and intelligence in their pursuit of highly personalized search experiences for their customers and users.
“The story of data scientists and search is a story of integration – and friction,” says Will Hayes, CEO, Lucidworks. “A lot of search platforms don’t have access to the libraries that data scientists have access to – this has required re-coding or translation of the models from Python into an operational pipeline. This requires a very specialized skill set that most data scientists don’t have. So operationalizing and testing models has been on the backs of developers. With Fusion 5.0, models can be serviced directly into search — eliminating a burdensome, human-intensive process.”
The Fusion 5.0 cloud-native architecture built on containers, microservices and APIs brings a new dimension of agility for data science tasks. Using the versatile Python SDK, models can be trained, modified, tested and published directly from Jupyter Notebooks. Fusion 5.0 provides all the flexibility organizations need for development with the governance and controls afforded by enterprise-grade production pipelines.
Lucidworks serves more than one-third of the US Fortune 100. Some of the world’s largest organizations, including REI, Department of Defense, and Staples.com already use Fusion to personalize the user experience. With Fusion 5.0, the handoffs and collaboration between data scientists, search developers and operations are becoming as seamless as they should be. Search is a team sport