Strategic information analysis is one of the most important activities that your company can perform. The fruits of this labor, especially when consolidated throughout the organization, inform everything from marketing and innovation, to risk management activities. To achieve this level of performance, you’ll need more than a simple monitoring or keyword recognition tool; you need software that reads and understands like an analyst.
Strategic information analysis is one of the most important activities that your company can perform. The fruits of this labor, especially when consolidated throughout the organization, inform everything from marketing and innovation, to risk management activities. To achieve this level of performance, you’ll need more than a simple monitoring or keyword recognition tool; you need software that reads and understands like an analyst.
Why is this important? Let’s look at a core element of information intelligence: Fact analysis. Determining what the facts are in a company’s claims and separating them from rumour or fiction, generally follows this process: 1) Identify the most relevant “claims” from the thousands of sources available; 2) Use as much contextual information as possible to evaluate its validity; 3) Collect the evidence and distribute the information to your organization.
More Read
For example, you’ll want to know when your competitors are coming out with a new product, shifting strategy, expanding to new markets or making important new hires. To do so, you will need to identify the relevant facts in multiple information types, from news and blogs to social media posts. Or, if you’re creating a risk profile for a new partner, you will want to integrate the D&B report with any recent claims linking them with risky or possibly illegal business practices.
Especially for text-based information, effective analysis requires a combination of human-level intelligence and contextual knowledge, along with the reach to access every possible piece of information. Given the volume and variety of data available today, this is simply impossible without the support of information analysis software.
So, what exactly should that software do? Ideally, it should read like you would, meaning that it should be able to identify the sections inside a piece of content that are relevant for your analysis, extracting specific sentences and claims, and present them to you for validation. Specifically, it requires these essential features:
a) Sentence-level categorization. If I am interested in monitoring innovation, I want to be able to specifically identify statements related to the topic “product announcement” or “patent assignment” that include the names of my competitors or of their leadership. Keep in mind that this could be written in a variety of ways, incorporating a range of writing styles, word choices and more that could be easily understood by a human–and should be replicated by a machine.
b) Ability to identify relationships between entities. The system should be able to understand whether there is a relevant connection between the event and the entity I am monitoring. If I am monitoring Company X and a news item states that Company Y lead by Mr. Smith, former VP of product at X, launched a new smartphone, the system should be smart enough to understand that Company X did not launch a new product, without human intervention. This functionality alone would significantly boost productivity by reducing false positives.
c) Extract entities and their attributes even if not included on the initial target list. Maintaining a continuously updated list of relevant employees of my competitor is a complicated task. However, this information is often available on the same sources I am already analyzing to identify relevant content for an event that I want to track. For example, in an announcement for a new product, “Mr. Smith, Vice President of Company Y since November” is quoted as saying that it will revolutionize the market; it later mentions that Mr. Smith attended Harvard. A smart information analysis software should learn from the analysis that Mr. Smith is also a target related to this event, and in the future, will trigger alerts.
d) Solve anaphora. In the example above, replace the second Mr. Smith with “he”. The system should understand that “he” is Mr. Smith, and therefore the announcement is relevant for your analysis.
e) Enable creation of a local knowledge base. Information analysis is an ongoing task.
Knowledge is built all the time, and it should be collected and stored in a form that can be easily shared among analysts. Don’t let your hard work go to waste!
