Claims:1. A System and method for the Psychological Stress Detection.
2. A System and method for the identification of Stress Level: as claimed in claim 1 involves following stages:
a) Collecting the Dataset from Social Media
b) Word Embedding
c) Glove Embedding
d) Building the LSTM – CNN model.
e) Training the Network
f) Testing
3. Collecting the Dataset In order to train our machine, we need a huge amount of data so that our model can learn from them by identifying out certain relations and common features related to the objects.
4. Building the proposed Psychological Stress Detection Model - This is most important entities considered. It consists of following-
1. Convolution Layer
2. Max Pooling Layer
3. Flatten Layer
4. Dense Layer
5. Dropout Layer
6. Activation Functions
5. A fully connected layer that takes the output of convolution/pooling and predicts the best label to describe the stress.
, Description:[1] The Depressive tweets and Sentiments of an individual user are gathered from twitter platform which is pre-processed and passed to the embedded CNN model which outputs user level attributes. Some of the libraries used in this application are Ftfy, nltk, re , keyed vectors, stopwords , porter stemmer etc. Ftfy is a module for making the text less broken which takes in bad Unicode and outputs good Unicode. The Natural Language Toolkit (NLTK) is an open source Python library for Natural Language Processing. It provides easy-to-use interfaces such as WordNet, along with a suite of text processing libraries for classification, tokenization, stemming, tagging, parsing, and semantic reasoning.

[2] A regular expression (or RE) specifies a set of strings that matches it, Regular expressions can contain both special and ordinary characters. Most ordinary characters, like 'A', 'a', or '0', are the simplest regular expressions. The special characters are ^,., $,* etc. In Keyed vectors mapping between words and vectors for: class: ~gensim.models.Word2Vec model. Used to perform operations on the vectors such as vector lookup, distance, similarity etc.

[3] Stop word is a commonly used word (such as “the”, “a”, “an”, “in”) that a search engine has been programmed to ignore, both when indexing entries for searching and when retrieving them as the result of a search query. The Porter stemming algorithm (or ‘Porter stemmer’) is a process for removing the commoner morphological and inflexional endings from words in English. Its main use is as part of a term normalisation process that is usually done when setting up Information Retrieval systems. Early stopping stops training when a monitored metric has stopped improving. Assuming the goal of training is to minimize the loss. Conv1D- ID convolution layer (e.g. temporal convolution).This layer creates a convolution kernel that is convolved with the layer input over a single spatial (or temporal) dimension to produce a tensor of outputs. Long Short-Term Memory layer is based on available runtime hardware and constraints, this layer will choose different implementations (cuDNN-based or pure-TensorFlow) to maximize the performance

[4] The Dropout layer randomly sets input units to 0 with a frequency of rate at each step during training time, which helps prevent overfitting. Inputs not set to 0 are scaled up by 1/(1 - rate) such that the sum over all inputs is unchanged. Max pooling operation for 1D temporal data down samples the input representation by taking the maximum value over the window defined by pool_size. The window is shifted by strides.

[5] The model takes in an input and then outputs a single number representing the probability that the tweet indicates depression. The model takes in each input sentence, replace it with it's embeddings, then run the new embedding vector through a convolutional layer. CNNs are excellent at learning spatial structure from data, the convolutional layer takes advantage of that and learn some structure from the sequential data then pass into a standard LSTM layer. The output of the LSTM layer is fed into a standard Dense model for prediction.

[6] The model is trained EPOCHS time, and Early Stopping argument is used to end training if the loss or accuracy don't improve within 3 epochs.

[7] We illustrate the performance of our proposed model achieved an F1-Score of 97.90%. Our proposed approach produces improved results by 4% comparing to the baseline approaches and is promising for the detection of stress level.