How is the customer service robot implemented? Dialogue recommendation system

This article is reproduced from the WeChat public account "Living in the Information Age". The author lives in the information age. To reprint this article, please contact the Living in the Information Age public account.

When you use many apps, you must have some understanding of the intelligent robot customer service system. Just like human customer service, customer service robots can have simple conversations with people and give corresponding answers to people's needs. Although the answers obtained most of the time are not very reliable, it is generally more labor-saving.

The recently popular chat robot ChatGPT is essentially a customer service robot, but the algorithm behind it is more sophisticated and the amount of pre-trained data is larger.

Let’s take a look at the technology behind customer service robots: dialogue recommendation system.

1. Overview of Dialogue Recommendation System

The process of users using the dialogue recommendation system is essentially a process that ultimately assists users in making decisions through multiple rounds of information interaction.

Conversational Recommendation System (CRS) breaks the information asymmetry barrier between the system and the user in the static recommendation system through rich interactive behaviors, allowing the recommendation system to Dynamically capture user preferences. One direction guides users to discover their new points of interest by exploring their current interests and preferences. On the other hand, during the interaction process, it accepts user feedback in real time, updates the strategy of the recommendation model, and achieves dynamic learning and updating. This is a recommendation-oriented dialogue system that captures user interests through online dialogue with users and recommends answers or products that users need.

General dialogue systems are usually divided into two categories: task-oriented and non-task-oriented. The latter is what people usually call chatbots. The task-oriented dialogue system is designed to help users complete specific tasks, such as helping users find the products they need, book hotels and restaurants, etc. Task-oriented dialogue systems for recommendation tasks can usually be regarded as dialogue recommendation systems that use natural language text and voice as the interactive form. In recommendation tasks, it has high commercial value.

2. Characteristics of dialogue recommendation tasks

From the application point of view of dialogue recommendation system, it has two typical characteristics: multi-round interaction and goal orientation.

1. Multiple rounds of interaction

In traditional systems, for example, when searching for products on Taobao, when users are looking for products with specific attributes, they will search them actively. conduct. For example, you can search for "men's jackets in spring". In this scenario, the user constructs the query himself. The recommendation effect not only depends on the search engine, but also relies more on the user's own professional knowledge to construct appropriate query keywords. This traditional recommendation system requires users to input possible attribute options based on their own prior knowledge in order to accurately locate appropriate products. But in many scenarios, users do not have such prior knowledge. In this case, users expect the system to proactively introduce potential items they might like to the user.

The multi-round interaction feature in the dialogue recommendation system can make up for the shortcomings of user active search in the traditional recommendation system. In the real-time interaction between the system and the user, it can show the user the unknown item attribute space by actively asking questions to the user, and use the user's feedback information to directly understand the user's needs and attitudes towards certain attributes, and build user interests. portrait to make correct recommendations.

2. Goal-oriented

The goal of the dialogue recommendation system is to recommend products that the user is interested in. Therefore, with the ultimate goal of achieving successful recommendation, we carry out In the interaction of obtaining user preference information, CRS and traditional recommendation systems have the same "recommendation" goal, but the two are completely different in the operation and implementation of the system. Traditional recommendation systems can be seen as the system unilaterally outputting recommended items to users. CRS, on the other hand, focuses on practical real-time feedback, constantly proactively explores user points of interest, and updates subsequent recommendation strategies.

3. Basic functional modules of dialogue recommendation system

A standard dialogue recommendation system consists of three functional modules: user intention understanding module, dialogue strategy module and recommendation module.

1. User intent understanding module

The user intention understanding module is a module that directly exchanges information with users. In the early years, its input was mainly dialogue text. With the development of technology, multi-modal data and user behavior data have increasingly become the main input of dialogue recommendation systems. Data Sources.

2. Dialogue strategy module

For recommendation systems, there is very little positive feedback data that can be based on, which creates a gap between the system and the user. The information does not match, and a failed exploration will waste the user's time, harm the user's preferences, and cause user churn. Therefore, pursuing the balance between exploration and gain is a key issue in conversational recommendation systems. The main task of the dialogue strategy module is to solve this problem.

In the process of multiple rounds of interaction, this problem is manifested in that the system needs to determine whether to continue to ask the user during the interaction process, or to recommend products based on the information that has been obtained, thereby increasing the user's choice of products. Probability. This is a typical game problem. Too many inquiries may cause user disgust, while too few inquiries may result in a lack of user preference information. Therefore, a good dialogue strategy needs to intelligently balance the two indicators of dialogue rounds and recommendation accuracy.

3. Recommendation module

The recommendation module is a module that implements the recommendation function in the dialogue recommendation system. Based on the user information that has been captured, it recommends the items that the user is currently most interested in. target item. In most CRS, the recommendation module uses a simple recommendation model, such as matrix decomposition. This is because a simple recommendation model can already meet the recommendation needs of the conversational recommendation system. Using an overly complex recommendation model will make the overall system complex. The degree increases, making the training of dialogue recommendation systems difficult.

The above is the detailed content of How is the customer service robot implemented? Dialogue recommendation system. For more information, please follow other related articles on the PHP Chinese website!