Deciphering Magic Leap’s History with NLP

Deciphering Magic Leap’s History with NLP

Companies lead interesting lives. They grow, contract, thrive or pivot, and their revenues are either increasing or they are counting on continuous funding to keep alive. They hire key people, then they lay off some of them, while they are sometimes the ones being left for good. They sue some company, only to be sued by others. They produce intellectual property hoping to be bought, or they buy other companies for their intellectual property.

Companies in all sizes and shapes experience innumerous events in their lifetime, but as a general rule, t is always very difficult to follow them. Startup databases such as CrunchBase or Pitchbook present lists of basic events such as funding rounds, key hirings, lists of investors and investments, along with a list of the recent relevant news articles. Yet, the information they provide is basic, and not particularly helpful for those who look for a higher resolution data on companies.

Digital company analysis startup Zirra has built an automatic process that delivers insightful outputs on private companies, including a list of possible competitors, their relative level of competition, web traffic, and an automated list of risk and success criteria, in addition to a map of meaningful events during a company’s life. Its proprietary technology allows it to create high-level analysis insights within seconds.

Among the meaningful events detected in the process are product launches, key people joining or leaving the company, mergers & acquisitions, legal issues, partnerships, and funding rounds. The main difference between Zirra’s platform and databases such as CrunchBase, CB Insights or Pitchbook is the use of technology to extract and analyze these meaningful events.

How does Zirra do it? Zirra collects and aggregates data from a myriad of public and private information sources, including structured directories, semi-structured databases, and completely unstructured text. It then utilizes Natural Language Processing (NLP) techniques to process this large volume of unstructured text and data.

After parsing out an article’s text, Zirra uses entity recognition to find mentions of companies in each article, and then link the article to each company that was found within. Currently, finding events is done using a pattern matching based approach, but machine learning trained models that identify events are just around the corner on the development chart.

After patterns are matched, a process of entity recognition on the matched portion begins so the event can linked to accordingly (i.e, WeWork is in partnership with Airbnb), also making sure that multiple links that need to identified as a single event are grouped together.

Zirra’s advanced capabilities in identifying meaningful events in a company’s lifetime is still in beta. However, it is the first and only service capable of categorizing forgotten or lesser known events, putting them in the right context and chronological order, thus helping investors, analysts, entrepreneurs, and business development managers track a company’s behavior both historically and in real time.

Let’s take AR tech company Magic Leap, who recently raised $502 million in series D, as an example. Searching for the company in Zirra’s homepage will generate the following timeline:

Source: Zirra.com

Let’s pick April 2017, a busy month for Magic Leap. As shown above, Zirra has detected a partnership between Magic Leap and CG animation studio Weta (event 1). Actually, there’s nothing new about the partnership, which goes all the way to April 2016, but a detection of a mention in the New Zealand press about the ongoing partnership tells us that the partnership is still active. This is an important sign in a world where partnerships can fall apart quietly.

Events 2,3, and 4 are actually connected. Zirra detected Magic Leap’s intention to buy Oscar-winning animation studio Moonbot (3), an event that resulted in hiring most of its artists and animators instead. As none of the founders joined Magic Leap (3), the algorithm detected officers leaving the company. At about the same time Magic Leap was rumored to buy Moonbot, it already had completed the acquisition FuzzyCube, a Texas-based game studio founded by former Apple employees (4).

Finally, the algorithm detected a departure from Magic Leap after BuzzFeed’s report about the lawsuit filed against Magic Leap by its former VP of strategic marketing, who alleged that she was fired after trying to correct the company’s gender imbalance. (5). The legal battle appeared later in Magic Leap’s timeline (see below).

Meaningful events detection is still a challenge, and Zirra’s product is still in beta. If you find any inaccuracies, please let us know. Detailed feedback is constructive, and welcomed, as meaningful events detection can only improve with your help. Please try it in Zirra’s homepage –  search for a company and scroll left to see its entire history. Click ‘view sources’ to see the origin of each event’s detection and let us know what you think.  

Comments

comments

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)
Loading...