Text Classification

Text Classification

Text classification is another feature made possible with Trendskout’s AI and its Deep Learning software. This service reads and interprets texts for a variety of applications in customer service, administration, sales, marketing and operations.

This interpretation of text is based on a text classification model that uses labeled text data for training. Just like other classification algorithms, it uses a two-stage model consisting of training and production, respectively. At the production stage, the previously trained model is used to classify and label new texts. The classification model is trained through different techniques, including neural networks and Natural Language Processing (or NLP, in short). Read on for a detailed explanation about the training and production stages.

Practical business applications

- Automation of admin tasks
- Reading and interpretation of texts and text snippets
- Workload reduction for customer service teams

How do organizations benefit from AI and Machine Learning?

Gain access to our Resource Center with interesting business cases and how-to's.
artificial intelligence

Powerful Cloud-AIOut-of-the-box with an intuitive interface, designed for non-data scientists

Text Classification in the AI Flow

1. Connect

2. Analysis

Text Classification

Text Classification is one of the available Trendskout AI Flow analysis-functions.

3. Automate

How does this work technically?

Step 1: Training the algorithm

In the first phase, labeled, or annotated, text data is uploaded in Trendskout. It is also possible to work with annotation via a text clustering step, which then assigns labels to training data. The training data can be uploaded via simple file upload, via API, plugin, database connection … in the connect step. The text data can be loaded in a raw format with practically no constraints on size or other properties. After all, as part of the algorithm selection & hyper-tuning phase, Trendskout will apply various operations to the raw text data in the training step. These operations are mainly NLP techniques, with the aim of cleansing this raw text data, semantically parsing it, and making it readable for the algorithm – a neural network – that will generate the predictive classification model. There are various options regarding data processing, algorithm selection, and parameters that lead to a virtually infinite number of possible combinations. Trendskout will use intelligent Auto ML & Solution Space exploration to find the most optimal combination and to generate the optimal model. The evaluation of the accuracy of this model is done by testing on a part of the training data that the algorithm has not previously processed. Hereby, among other things, avoiding overfitting. The entire training process is started – as with the other analyzes – by clicking on the “Train / Deploy” button in the AI ​​Flow interface.

text classification

Step 2: (Real-time) Text classification

The winning classification model from the training step is then used to classify or label new texts. Bringing the classification model into production takes place – as with the other analyzes – by clicking on the “deploy” function. There are various input options for these texts, API, Plugin, Database, FTP etc. The texts can be uploaded in batch or real-time. Due to the high processing speed of the texts, NLP and classification, the answer always follows in real time. This answer is a label, or classification, in accordance with the annotations from the training data.

Text Classification + Trendskout

Text classification can be used in an AI flow via the accessible visual drag & drop Trendskout interface. In addition to linking data input, automation actions can also be added such as integrating with an external system via API or plugin, writing to a database, sending out parameterized e-mails … Just like with other AI and Deep Learning functions, the models are continuously evaluated for accuracy, also after the initial training phase. In this way, the most optimal model is always used for your text classification. Even when the properties of your documents, comments or other types of text change.

Ready to discover all features during a live demo?Get in touch and we will be happy to show you the direct business value of artificial intelligence for your organisation.