NLP : Rediscovering Happiness

Do you often feel depressed? Want to talk to someone who can make you feel better? How about going to a psychotherapist? Or do you feel shy in going because of social and mental factors? 

We often feel sad, moody or low from time to time. Some people experience these feelings intensely for long periods of time and sometimes without any apparent reason. Depression is not just a low mood – it’s a serious condition that affects your physical and mental health. 1 in 4 worldwide will suffer from a mental health condition in their lifetime. A person with a “learned pessimism” style of thinking always gets depressed. This pessimistic thinking can come due to various reasons – family problems, medical illness, drug and alcohol use and personality etc.

 So how about having a counsellor with you that tries to get more insights about your problem and give you techniques that can help you in stress.

Strategies of Depression:

Psychologists have formulated down the below mentioned strategies which can bring negativity to a person’s mind:-

1.     Chunk Up in Problem Situation: Imagine a situation where a person applies for a job. A person with a negative thinking would say, “No one would employ me.” Or “Why do I always do something wrong?” while a person with temporary explanatory style thinking might say, “I hadn’t allowed myself the time to prepare well enough for the interview.” The person who gets depressed mainly thinks about problems, so they chunk up in their goals setting unrealistic outcomes for day to day problems.

2.     Inclination towards the Past:  The depressed person is always stuck at his past. He never tries to move forward. In his present situations he brings instances from his past for e.g. “I know why I am unhappy; it’s because I never got over my breakup.”

3.      Negative Self Talk: A person differs from the other in the amount they pay attention to their internal self talk. Rumination is a deep thought about something. Rumination has no problem in itself but when combined with permanent pessimistic thinking creates problems. They sense a feeling of helplessness all the time. The more they try to ruminate, the more it arises and the more depressed the person will feel for e.g. If a person ruminates on something upsetting his close ones has done, he may take longer time to forget and in worst case may try to hold a grudge and might even destroy a relationship.

4.     Lack Of Sense Of Control:  A person who is often depressed lack a sense of himself in making decisions. Whenever they engage in an activity they mainly think of the output coming as a bad result for e.g. “If I take up this task I may not be able to finish it and may have bad consequences.” Or “What is the point of doing this task when you know it won’t terminate in the way it should have been.”

NLP Approach:

By looking at the above strategies we get to know that the depression arises by the beliefs that things can’t change. The key to heal depression is to make positive thoughts arouse in person’s mind, to make them look ahead in future and not let them think too much about something.

Using NLP we can parse, tag and extract information from. It can be used in question-answering mechanism. To answer questions related to mental problems or to ask questions based on user’s problems; NLP model must be applied to some training data to analyze the linguistics aspects of conversations between a person and a counsellor.

Experiments and Analysis:

Researchers from Stanford University conducted an experiment on counselling conversation strategies that can be put to use for an application of NLP. They collected data from an SMS texting-based counselling service where people in depression engage in therapeutic conversations with counsellors. The data consisted of millions of messages from 80,000 conversations with counsellors.   

They analyzed that to make NLP application powerful in dealing with the depressed person following actionable strategies must be looked into:-

1.     Adaptability: Measuring distance between vector representation of the language used in conversations going well and going badly. Good counsellors tend to be more sensitive while seeing the trajectory of the conversation and they react to it accordingly.

2.     Dealing with Ambiguity: A clustering method can be developed to measure differences in how counsellors respond to similar ambiguous situations.

3.     Creativity: Good Counsellors try to answer in a more creative way by not giving too generic reply for e.g. Rather than asking “Would no one employ you?” the counsellor should focus on counter-example “When has there been a time that someone did employ you?”

4.     Making Progress:  A sequence-based unsupervised conversation model can be developed to discover ordered conversation stages common to all conversations. After the conversation ends, counsellor may ask a follow-up question “How are you feeling now?” which can be used to compute the efficiency of the application and can improve future conversations.

Challenges:

·        It is difficult to monitor the behavioural changes of the depressed person continuously throughout the conversation.         

·        Sometimes it may happen that the same response may not fit in every situation. So, a model must be powerful enough to handle situations like these.

Real Life Systems:

·        IBM Watson is a question answering system developed by IBM that can answer questions posted in natural language. It is used in decision making in lung-cancer treatment at Memorial Sloan Kettering Cancer Centre, New York.

·        Siri is a personal assistant developed by Apple. It uses voice queries and a natural language UI to answer questions, make recommendations and perform actions by delegating request to Internet.

Considering how complex these systems are to build, NLP Models can not replace rather they will assist diagnosticians.

References:

1.     Natural Language Processing in Mental Health Applications using non-clinical texts - Rafaela. Calvo, David N Milne, Sazzad M. Hussain and Helench Christen Sen.

2.     American Psychiatric Association Diagnostic Criteria from DSM-IV TM, American Psychiatric Association, Washington DC, 1994.

3.     Large Scale Analysis Of Counselling  Conversations: An application of Natural Language Processing to Mental Health – Tim Althoff, Kevin Clark, Jure Leskovec.

4.     Karthik Dinakar, Allison J.B. Chaney, Henry Lieberman, and David M. Blei. 2014a. Real-time topic models for crisis counselling in KDD DSSG Workshop.

5.     http://www.transformations.net.nz/trancescript/nlp-and-the-rediscovery-of-happiness.html