Sequential Recommendation Model Training for Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sequential recommendation systems fail to adequately utilize information about document subjects in training data and live recommendations, leading to inaccurate predictions of user actions.
Innovation Solution
A novel sequential recommendation system that trains a machine learning model on document information and user actions to generate more accurate predictions, using a two-stage training process to learn document vocabulary and recognize user action sequences, allowing for predictions of next user actions based on input sequences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing sequential recommendation systems are used, then the system structure is simple, but the prediction accuracy is poor due to inadequate utilization of document subject information
Solution Approach 1:
The training process is segmented into two distinct stages: a first stage that trains the model to recognize document subjects and a second stage that trains the model to recognize user action sequences. This segmentation allows each stage to focus on specific aspects of the data, improving overall prediction accuracy without overwhelming complexity in a single training process.
Solution Approach 2:
The system performs preliminary training on document subjects before training on user action sequences. By pre-training the model to recognize and represent document subjects, the system prepares the model to better understand and predict user actions, improving prediction accuracy through preparatory learning.
2Measurement precision
If the model is trained on comprehensive document information, then the prediction accuracy improves, but the training time and computational resources increase
Solution Approach 1:
The training process is divided into sequential stages where the model first learns document subject representations and then learns user action sequence patterns. This segmentation allows the model to build up knowledge incrementally, improving prediction accuracy while managing training time through structured learning progression.
Solution Approach 2:
The system performs preliminary training on document information to establish a foundation of subject representations before training on user actions. This preliminary action reduces the computational burden during subsequent action sequence training, as the model already has established document understanding capabilities.
3Measurement precision
If the system utilizes document content information, then the prediction accuracy for next user actions improves, but the complexity of data processing increases
Solution Approach 1:
The data processing is segmented into two phases: first processing document content to extract subject information and train the model to recognize subjects, then processing user action sequences to train the model to predict actions. This segmentation reduces overall processing complexity by handling different data types in dedicated phases.
Solution Approach 2:
The model's intermediate representation of document subjects acts as a mediator between raw document content and user action predictions. By transforming document content into standardized subject representations, the system simplifies the connection between document information and action prediction, reducing data processing complexity.
Data Source
AI summary
A method for predicting a next user selection in an electronic user interface, such as a website, includes training a machine learning model according to a training data set, the training data set including a plurality of token sets, each token set representative of a respective document accessible through the interface, each token set including a plurality of words, each word describing a characteristic of the document, to create a trained model. The method further includes receiving, from a user, a sequence of selections of documents, inputting the sequence of selections to the trained model, and outputting to the user, in response to the sequence of selections, a predicted next document selection according to an output of the trained model.