It is frustrating to spend time looking for information when you could be completing a task.
Search has become ingrained as part of everyday life.
Pre-Internet Findability
Today, there’s no need to pull a volume of an encyclopedia off a shelf or even leave the room to find answers to questions. One can simply use phone to search for answers to questions. Google and Wikipedia have redefined what it means to search. But have they made search any more intelligent? They certainly satisfy the itch to correct people on event dates, geography, and historical characters.
When it comes to the workplace, however, search encompasses a great deal more than fact checking, and intelligent search goes well beyond the web.
Search has gone mainstream. People use the word “search” when they want to locate a retail store or book a hotel. That simplistic notion of search does not carry over particularly well to finding information essential to doing your job.
Teasing Out the Meaning of the Search
Part of moving from a simple Google search to a more sophisticated model involves language.
Standardizing content in one format—her example is high-definition PDFs—creates better visibility and fewer irrelevant search results. You may be able to avoid overly complex algorithmically based search engines by improving content processing, eliminating duplication, and using a single taxonomy.
Use better metadata and better data.
Almost anyone looking at search within the enterprise stresses findability. If you’re looking for the company’s holiday schedule, you don’t want the one from 3 years ago, you want the most recent one.
Similarly, if you are building a web site for external use, you want potential customers to find what you are selling. You want to back up your sales efforts with excellent customer service. This is another opportunity for intelligent search, since customers increasingly prefer to help themselves without using an intermediary. They like self-service, but only if it answers their questions.
Semantics plays a role in customer service. Its analysis of the contextual meaning of words enhances the quality of answers. For example: customers might enter “How much will it cost me…” while your search engine understands phrases as “What is the price…” To be findable, your customer’s search query must translate to your words. Synonyms dictionary would help to resolve this issue.
Definition of intelligent search goes beyond findability. A search engine should know what you need and what your colleagues found valuable, and supply it to you when you need it.
For Coveo search engine, the power and sophistication of machine learning technology is the driving force behind intelligent search. Intelligence springs from usage and analytics data, along with a multitude of other factors, the components of which are hosted and managed by companies such as Coveo.
Regardless of how you define intelligent search, it’s clear that enterprise search requirements go well beyond what Google or Wikipedia can provide. Different approaches to intelligent search provide much to think about when implementing, redesigning, and rethinking enterprise search. Intelligent search goes well beyond what searching the web looks like.
Improving Search and Decision-Making with Semantics
We’ve all heard about how Google’s proverbially simple search form has led professionals to expect similar simplicity from search solutions provided by corporate IT. Except this model doesn’t really work, and it’s costing millions of dollars every year in time wasted when professionals don’t find, and have to re-create information.
The reason it doesn’t work is that while every organization has a specific worldview, search engines are essentially blind. Worldview is the inventory of business objects that an organization cares about (products, geographies, customers, processes, etc.) and their relationships, that are typically captured in a taxonomy or ontology. While professionals implicitly want to search for information according to their worldview, search engines don’t offer them a practical way to do so.
Semantics Provides Meaning
The missing piece in this puzzle is a “meaning engine” that would understand unstructured content through the lens of your organization’s worldview. It exists: it’s called a semantic enrichment platform.
A semantic enrichment platform ingests your organization’s taxonomy or ontology and applies it to your content at scale. Leveraging natural language processing, it understands your content the same way humans do. It recognizes topics that are relevant to your business, entities of interest, their attributes and relationships, and converts them into structured data, that can be used standalone, or as metadata describing your content deeply and consistently. In energy, for example, entities of interest might include commodities, trading companies, and the countries where they do business.
Better Metadata Accelerates Search
When used as metadata, this data acts as an eye-opener for search engines that can finally see your content through your own worldview. This redefines the search experience by offering end-users new tools to locate what they are looking for.
Faceted Navigation enables end-users to search by business entity or topic (for example by company name, commodity type or region), helping to find the most relevant content in just a few clicks.
Links to relevant information provide convenient access to structured information about entities of interest so users don’t have to collate it themselves. For example, each company name could be linked to data about its activities. Topic pages concentrate all information about a specific topic in one convenient access point so users don’t need to sift through all other materials to access it. A topic page on electricity would, for example, filter out information related to other energy sources.
Content recommendation uses metadata to surface other documents with similar topics, promoting discovery of relevant information. A document on a merger in the gas sector might point to reports of other, similar operations.
Such mechanisms significantly accelerate and simplify search tasks, offering not only time and cost savings, but also more informed decision-making.
Better Data Improves Decision-Making
But semantically-extracted information can be used for its informative, rather than descriptive, value. Not as metadata, but as standalone data. This opens the door to applications that address the above blindness at a deeper level, providing higher-level and faster insight into the subject matter at hand.
One of semantics’ capabilities is to recognize not only entities, but also their relationships (often expressed as triples). One such relationship might for example indicate that company A is a “supplier of” company B. Information value from these relationships may come into play under a variety of scenarios.
Knowledge Bases (or Graphs) integrate such structured information at scale so they can then be queried. One might contain, for a given commodity, links to all suppliers.
Complex Reasoning can be performed on these knowledge bases, enabling business applications to provide higher degrees of automation in decision-making tasks, for example, automatically balancing supply by identifying alternative suppliers when one announces production issues.
Analytics and Visualizations provide dashboards that sit on top of the data and reveal its meaning on a more holistic level. For example, a network graph could plot all company relationships in natural gas, indicating which companies might be exposed to increasing prices in a given region.
Lastly, semantics can also be used to deliver Question Answering Systems that offer users a way to get answers to questions formulated in natural language (“Which electricity providers have the most diversified supply chain?”) instead of engaging in search.
Semantics Provides Faster Insights and Better Decisions
As can be seen from the examples above, semantics is the “meaning engine” that ensures that users can overcome search’s blindness and access information through the specific worldview relevant to their work. But this engine brings meaning to more than your search engine: it is your information management as a whole that benefits, bringing the promise of smarter applications that efficiently handle more of the groundwork, accelerate time-to-insight and support better decisions.
Self-learning
Intelligent search is no longer a nice-to-have feature in organizational information systems; it is a critical part of how businesses are transforming the way they work. Intelligent search goes beyond findability and information access. Like a trusted advisor, intelligent search knows what documents you need for your tasks and which articles your colleagues found most valuable and would be useful to you too, and simply gives everyone the information they need, when they need it. And the power and sophistication of machine-learning technology is the driving force behind intelligent search.
What Is Machine Learning?
Machine learning learns from and makes predictions on data. Applied to search, every time a user performs an action on your web site or support portal, he or she provides data about what is useful. Did they submit a support ticket? That means the articles they just read did not help. Do most people spend only one minute with a document that would normally take 10 minutes to read? That’s a sign that the content isn’t useful, or perhaps it’s too difficult to understand. With machine learning, all of that information and more can be used to make data-driven predictions and decisions without manual intervention.
How Will Machine Learning Make Search Intelligent?
When someone submits a search query or clicks on the third search result, they are implicitly telling you what is most relevant. As your online community members download content, visit various web pages, watch videos, start an online chat with your support agents or submit support tickets, their behavior provides information on the relevance of the content they come across. This behavioral data as well as search behavior which signals intent are captured by search usage analytics.
Intelligent self-learning search engines powered by machine learning can leverage such usage analytics data to continuously self-learn. This improves search relevance and hence, the self-service experience on your community in many ways. For example, automatic fine-tuning and ranking of search results based on machine-generated predictions about what is most useful improves the experience of all community members.
Without machine learning and analytics data, administrators need to fine-tune search rankings manually: create boosting rules, add synonyms, promote documents, etc. Because relevance is an ever-evolving process and the document that was the most relevant last week may no longer be relevant today, it is almost impossible for administrators, especially those at large organizations or those with multiple product lines, to keep pace with the rate of change.
With machine learning, highly manual and complex enterprise search can be transformed into intelligent, self-learning and self-tuning search.
Why Now?
Machine learning has been around for a long time. It used to be very complex to deploy and manage. Collecting usage data, managing databases, provisioning servers, developing and maintaining machine learning algorithms and using machine-learning predictions in the search system were typically very complex. This would require data scientists, database experts and developers. Only the biggest organizations could afford that. But the fast adoption of cloud solutions has made the use of machine learning much easier, cheaper and more attainable. In particular, the recent trend towards cloud-based enterprise search is a game changer.
What Is the Impact of Cloud-Based, Self-Learning Search?
With cloud-based, self-learning search, all the required components are hosted and managed by the vendor, such as Coveo. Because of its scalability, it has the potential to change the customer service industry the same way machine learning has impacted e-commerce and social networks.
In the past, the high cost of using and managing machine-learning systems meant that machine learning was rarely used for traditional enterprise search or self-service support sites. The cloud makes that affordable to all customers and to all departments, especially when deploying self-learning search on self-service support sites and on communities, because of its ability to scale and handle large volumes of data.
