There are many challenges in sentiment analysis. 
1. The use of creative and non-standard language in text can be difficult to interpret by the automatic natural language systems.
2. Utterances sometimes can convey more than one emotion or refer to emotional events without explicitly or implicitly to express the emotions/feelings of the author of the text.
3. Text that were used in different contexts and different senses can sometimes express different feelings/emotions.
4. There might also be a lack of paralinguistic information or lack of large amounts of labeled data that will be difficult for many machine learning algorithms to understand.

Challenges in Sentiment Analysis

1. Evaluation (good or bad)
2. Potency (strong or weak)
3. Activity (active or passive)

Dimensions of meaning of Sentiment Analysis of Text

Many of the machine learning approaches for automatic detection of sentiment are supervised at first. In other words, the systems first learn a model from a set of example instances labeled with the correct sentiment also known as the training set. Then the model is able to predict the sentiment of new instances. 

This paper also discusses the many angles that we can use sentiment analysis, for instances, we can detect the sentiment of the writer or reader, detect sentiment from different the words, sentence, paragraph level, and it can also detect the sentiment towards a target.

Process of Sentiment Analysis

Currently, sentiment analysis is used in many areas of our lives. For example, sentiment analysis is used in politics for tracking the public sentiment especially in social media, and it is also used in public health to detect emotions that might signals depression and tracking well being. Besides these, it is also used in brand management, stock market, and education.

Sentiment Analysis areas

Currently, sentiment analysis systems have been applied to many different kinds of texts including novels, news paper headlines, emails, etc. These systems can detect the subjectivity, valence, and emotions in sentences and tweets.

Detecting Subjectivity, Valence, and Emotions in Sentences and Tweets

In almost every language, the same word can often convey different sentiments in different contexts. For example, the word unpredictable is negative in the context of automobile steering, but positive in the context of a movie script. However, the majority of these words tend to convey the same sentiment in a large majority of the contexts they occur in. For instance, great and amazing are positive, and sad and depressed are negative in most cases when they are used. These majority associations are often referred to as prior associations, which is beneficial for sentiment analysis systems (where the system creates lists of term–sentiment associations by manual annotation).

Capturing Term-Sentiment Associations

Sentiment analysis is used to refer to the task of automatically determining the whether a piece of text is positive, negative, or neutral. It can also determine someone's attitude towards a particular target or topic.

University of Michigan - Ann Arbor

The categorization of texts into organized groups, which is usually implemented by supervised learning.
Applications: sentiment analysis, spam detection, news categorization, topic analysis, and more.

Text Classification

One of the earliest linear embedding models, trained on 200,000 words using less than 1 million parameters.

Latent Semantic Analysis (LSA)

Graph neural networks capture the internal graph structures of natural language, like syntactic and semantic parse trees.

Graph Neural Networks

A text classification category that views text as a "bag of words." This is the simplest model for text classification, but is still very successful. 

Steps:
1) Learn the vector representation for each word using an embedding model
2) Take vector sum or average of embeddings as the text representation
3) Pass through one or more feed-forward layers, known as Multi-Layer Perceptrons (MLPs)
4) Perform classification on final layer's representation using a classifier

Feed-Forward Neural Networks

Siamese Neural Networks are a special case of text classification designed for text matching.

Siamese Neural Networks

Linear classifiers use a linear combination of inputs to predict the category.

Linear Classifiers

Sentiment Analysis

Text-pair classification extends standard classification methods to process two distinct texts simultaneously. When provided with a pair of texts composed of tokens $$x_1 \dots x_m$$ and $$y_1 \dots y_n$$, the two sequences are concatenated into a single combined sequence. The total length of this unified sequence is $$len$$, calculated as $$len = n + m + 2$$ to account for necessary special tokens. A classification label is subsequently predicted for the entire sequence by utilizing the aggregated representation vector, specifically $$\mathbf{h}_{\mathrm{cls}}$$. This overarching framework addresses multiple NLP challenges, including semantic equivalence judgement (assessing if two texts share identical meanings), text entailment judgement (evaluating if a hypothesis logically stems from a premise), grounded commonsense inference (gauging the likelihood of an event given its context), and question-answering inference (verifying if an answer correctly matches a question).

Text-Pair Classification

In text classification, a specific category or label is assigned to a piece of text. For example, in the context of sentiment analysis, the sentence 'We had a delightful dinner together.' would be assigned the label 'Positive'.

Learn Before

Related

Learn After