Recommender Systems

This week we learnt about how users of similar taste in a product or interests may be grouped and make recommendations accordingly. Such consumers have the potential to support and improve the quality of the decisions consumers make while searching and selecting a product.

There are two main approaches to recommending:

collaborative filtering

content-based

However, other approaches i.e hybrid also exist. 

For collaborative filtering, the recommendation is based on prior user behavior. It may exist based solely on a single user's behavior or form the behavior of users who have similar traits which may come out more effectively.

This system is applicable in many areas and an example that stood out for me would be in the event of shopping online. If several users have bought a product online and have liked that item, the are then grouped and the recommender system is able to suggest similar products to these users that they may like. 

There is the assumption that these users rate the product honestly and those who've had a similar taste in the past will have a similar taste in the future.

However, this method is faced by few challenges such as:

cold start - whereby there needs to be enough users for the system to find a match.

sparsity - finding many users of the same taste may be difficult despite the number of users.

first rater - a product cannot be recommended if it has not been previously rated.

popularity bias - a user with unique taste cannot be recommended for.

As for Content-based filtering, recommendations are based on information of the content rather than users' opinions. If a user loves to read about sports on the newspaper, the system uses this information to recommend more sports articles for the user. This system is advantageous because it can recommend to users with unique tastes.

Hybrid approaches combine both collaborative and content based filtering which effectively increases the efficiency of recommender systems.

Verma, Y.V. (2021).

Recommender systems are useful for both the value for the customer and value for the provider.

As for value for the customer, the system allows:

    Finding Interesting Things: delivering goods, services, or experiences that satisfy the needs or arouse the curiosity of the client.

    Narrowing Down Choices: Helping clients save time and effort by sifting through a multitude of possibilities to choose what best meets their needs and preferences.

    Providing a wide array of options for clients to consider can enable them to either pursue new interests or hone their current ones. This approach is known as "Exploring the Space of Options."

    Finding New Things: Introducing clients to unique products, ideas, or experiences that they might not have come across previously, broadening their perspectives and encouraging a spirit of exploration.

    Entertainment: Improving the whole customer experience by offering pleasure and entertainment throughout the process of investigation and discovery.

For the value of the provider, the recommender system facilitates for:

    Extra Customized Service: Differentiating a provider from rivals by providing customized experiences and services can boost client happiness and loyalty.

Enhanced Trust and Customer Loyalty: Providers can foster relationships and establish trust with their clientele by offering customized services and paying attention to each individual's demands. This will result in repeat business and long-term loyalty.


    Enhanced Sales, Click-Through Rates, and Conversions: Tailored offerings and focused marketing campaigns have the potential to elevate customer engagement, which in turn boosts important performance indicators like sales, click-through rates, and conversion rates.

    Possibilities for Persuasion and Promotion: By utilizing customized interactions and services, businesses can increase the efficacy of their marketing and sales initiatives by promoting particular goods and services that are catered to the interests and preferences of individual customers.

    Acquiring More Information About consumers: Providers can improve their offerings and marketing tactics by learning more about the requirements, preferences, and behaviors of their consumers through personalized encounters that yield data and insights.

To summarize, businesses can establish a mutually beneficial connection with their customers by prioritizing value for both the supplier and the client. This can result in tailored services, heightened trust, and focused promotions that ultimately boost customer happiness, loyalty, and engagement. For the supplier, this results in increased revenue, click-through rates, and conversions in addition to offering insightful data about the preferences and actions of the clientele. Businesses can attain competitiveness in the market and sustainable growth through this reciprocal exchange of value.