Time is one of our most valued assets in life. Everyone always wants more time. One of the biggest reasons people don’t like to answer surveys is because it takes up too much of their time.
SurveyMonkey 🐵, is a company built around getting you the answers you need. But, in a recent survey conducted by SurveyMonkey themselves, they found that 60% of people don’t want to take a survey longer than 10 minutes and 87% of people don’t want to take a survey longer than 20 minutes.
Furthermore, a survey conducted by SurveyAnyplace found that in-person 🧔 surveys had a 57% response rate while online surveys 🖥️ only had a 29% response rate. This could be due to the personal aspect of the in-person interview 🗪 or due to the fact that the client could just speak instead of typing everything they had to say. However, both these problems could be solved using natural language processing.
Whether it be 10 or 20 minutes, most people find that they don’t have time to open up their computers and answer your questions. But, what if there was a way for people to answer your survey while doing other important tasks. Enter artificial intelligence (AI) 🧠 and natural language processing (NLP). NLP gives technology the ability to speak and understand whichever language you speak.
How does NLP work?
Natural Language processing can be extremely complex, so it’s broken down into 4 easy steps:
- Tokenization 💰
- Part-of-Speech-Tagging 🏷️
- Parsing/Tree diagrams 🌳
- Semantics 💬
Tokenization 💰
Tokenization is the first step in NLP. What tokenization does, is it divides the sentences into its different parts. Essentially, it takes a sentence and it breaks it up into the words that form it. Pretty easy right ☺️.
Part-of-Speech-Tagging 🏷️
The second step of NLP is part-of-speech-tagging. In the English language, there are 9 fundamental types of words such as nouns, adjectives, and verbs, etc. And knowing which type a word is, is essential for the computer to figure out what the sentence is saying. However, due to the ambiguity of words in the English language, your computer can’t tell us what each type of word means. For example, “leaves” can be a noun or a verb. So, to resolve this problem, we look at the next step in the process.
Parsing/Tree diagrams 🌳
The third step in NLP involves looking at phrase structure rules to determine the type of word. For example, a rule says that a sentence can be comprised of a Noun Phrase followed by a Verb Phrase. Another rule states, a noun phrase can be made by an article (like “the”) followed by a noun or an adjective followed by a noun.
Then, if you compile all of these rules into your computer, it can help you identify the types of words in each sentence which will help you create your tree diagram. The tree diagram also known as a parse tree can tell you how the sentence is constructed and tells you which words are important in the sentence.
Semantics 💬
Finally, the last step is semantics. Being able to understand what type of word is which is important, but being able to tell the meaning of the word just as important. In the English language, there is also ambiguity in the meaning of a word, even if the word types are the same. For example, “board” can be a chopping board or a surfboard. In most cases, NLP solves this by looking at the words in front of the ambiguous word or those after it.
So, if we were to say “We were cutting potatoes on the board,” we know the board we are referring to is a chopping board since potatoes 🥔 precede the word “board”. Whereas if we said, “I’m hitting some big waves with my board,” we know the word board refers to a surfboard 🏄 as it the word waves precede “board”.
With these 4 steps, computers are able to understand what we are saying to them without too much confusion. However, there is still work to be done, with challenges like irony and sarcasm, there is still a lot of room for improvement when it comes to NLP and understanding what humans are really trying to say.
Real-Life Applications
This is the technology that Siri, Cortana, Alexa, and other smart assistants use to understand what you are saying and give answers to your questions. With AI and NLP, you can use Siri or your other smart assistants to ask you the questions in the survey. Then you can answer using your voice, and whatever you say will be recorded and sent back to the survey maker by your smart assistant.
The NLP technology would allow you to have more of a personal conversation, as it feels more like you are talking to someone rather than answering questions on your computer. Furthermore, this would allow you to do other things while answering the survey.
Instead of being confined to your computer 💻 and having to answer it on your laptop, you could be cooking, cleaning, or going for a walk while answering the survey. This could bring in more responses in surveys and if SurveyMonkey were to implement this, it could change their response rates drastically. In turn, this can help them help their customers improve their businesses so everybody wins.
Key Take-Aways
- NLP functions in 5 steps
- Tokenization 💰
- Part-of-Speech-Tagging 🏷️
- Parsing/Tree diagrams 🌳
- Semantics 💬
- Lots of technology like Siri and Cortana already use NLP
- Surveymonkey 🐵 can use NLP in the future to improve their services and encourage survey completion
