Search Technology: Greater power, but increasingly simple to use
Adding machine learning
In the last few years, Coveo has focused intensively on making machine learning more accessible to organizations of all sizes. “We built several machine learning models for various uses cases,” said Nep Vijayaraja, product marketing manager at Coveo. “Our customers can go on their Coveo web console to configure the model they need for their use case. Most companies don’t have data scientists available, so we do the work of maintaining the machine learning models.”
Coveo can provide a highly personalized experience because it can integrate diverse information into a meaningful pattern. “If someone orders boots online, and lives in Montreal, and it is February,” says Vijayaraja, “they should not receive the same purchase options as someone who lives in California. The same goes for accessories. In one case, the recommendation engine might offer gloves, while in the other, a summer hat would be more appropriate.” Coveo is able to give any number of individuals an entirely unique experience, because of its ability to scale rather than having to segment the data into aggregated groups of customers in order to reduce the number of cases.
Coveo Machine Learning understands the user’s intent based on their context. “If the individual is logged onto an ecommerce site or is using an intranet, Coveo Machine Learning has another level of contextual information, since we know exactly who the person is,” said Vijayaraja, “This allows Coveo to suggest what the user needs at the moment, and even what they are likely to need next.”
Leveraging search for compliance
The transition from search engine to insight engine has been marked by increasing capabilities in not just finding information but also interpreting it. The financial services industry has undergone numerous changes in the last decade, including greater regulation and greater awareness of various types of risks. Sinequa has made significant strides in this market with its Insight Platform, which has multiple features that make it a good match for these applications.
One of the world’s top 20 banks wanted to be able to identify confidential data more consistently and efficiently. “Their existing process relied on their employees manually evaluating and tagging the confidentiality level of a document,” said Scott Parker, director of product marketing at Sinequa. “The Sinequa implementation automated and accelerated the identification and categorization of confidential data using advanced natural language processing. Then, we built and trained a machine learning model to predict document confidentiality more precisely,” continued Parker. “This resulted in the bank saving millions of dollars, while automating compliance.”
Diverse data sources
Sinequa is well-suited to financial services applications because it can work across the extreme diversity of data sources found in the industry. These include in-house legacy systems, cloud platforms, structured and unstructured data, which can be presented as a cohesive view to the end user. “The user might want to prepare for a client meeting and need the most up-to-date fund information,” Parker said, “or integrate a post-merger acquisition quickly and get insights across both the acquiring and acquired company systems. Sinequa’s Insight Platform can provide answers to all these questions and deliver it proactively.”
The text mining and semantic capabilities of Sinequa allow it to be more adaptive in tasks such as routing service tickets. In a traditional workflow, the routing path would need to be established in advance, so changes in products or customers could be difficult to make. With Sinequa’s search and text analytics, the routing path could be adjusted more easily because it is accommodating to new content as it comes in.
Enterprise search has often been criticized by users as ineffective, but the nature of language poses significant challenges. “Establishing context is difficult for computers,” Parker commented. “The searcher might have an intent in mind, but cannot articulate it unambiguously simply because language has a lot of subtleties. In addition, language is full of idioms. Sinequa has an advantage on this front, because it began as a linguistic product. Our software can adapt to new buzzwords and content as they emerge.”