Named entity recognition simplifies data entry by allowing users to enter sentences instead of filling out specific fields. However, there is a risk of translation algorithm failure if the program cannot understand variations in phrasing.
While from an end user’s perspective the process of entering data into a computer is quite simple, the process actually starts many other smaller processes. For each piece of data, the computer has to translate that information into a language it understands. Computers use an internal language known as binary in which a collection of “1” and “0” characters is formed to tell the computer what to do. Named entity recognition is a method of streamlining the process of translating from input to data by helping the computer break down sentences into their component parts.
When you enter data using most computer programs or web pages, the program or web page usually prompts you to enter specific information in each field, such as the “Name” or “Address” fields. A named entity recognition program eliminates most of that, allowing the end user to enter strings – sentences – of text instead, providing a more natural interface. The program takes the data you entered and sorts it into parts that it can understand more easily.
Take a phrase like “Joe ordered four bushels of bananas.” In a traditional program, the end user would enter “Joe” in the “Employee Name” field, “Bananas” in the “Type of Purchase” field, “four” in the “Quantity” field, and “bushes” in the “Units” field . In a named entity recognition setup, the end user simply enters the entire phrase as written. As soon as the user clicks Submit, the computer sorts the sentence into parts, performing the same data splitting that the user would have used previously.
While this is much leaner than a traditional input program designed by fields, there is one major drawback: the potential failure of the translation algorithm. Less intelligent named entity recognition algorithms might only “understand” the phrase in a specific way. In other words, if the employee rephrases the sentence and writes, “Four bushels of bananas were ordered by Joe,” the program may have difficulty understanding and ordering the parts. For this reason, even if the end user is no longer responsible for entering data into the fields, he is still not as free as the system might seem at first glance.
Protect your devices with Threat Protection by NordVPN
