Browse Prior Art Database

Method and System for Tracking Performance of a Brand

IP.com Disclosure Number: IPCOM000237246D
Publication Date: 2014-Jun-10
Document File: 5 page(s) / 231K

Publishing Venue

The IP.com Prior Art Database

Related People

Pengyuan Wang: INVENTOR

Abstract

A method and system is disclosed for tracking performance of a brand, a product category, or an online network. The method and system provides a unified framework to combine multiple signals to estimate a performance indicator for one or more of, brands, categories and networks. The method and system uses a Principal Component Analysis (PCA) model for computing the performance indicator.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Method and System for Tracking Performance of a Brand

Abstract

A method and system is disclosed for tracking performance of a brand, a product category, or an online network.  The method and system provides a unified framework to combine multiple signals to estimate a performance indicator for one or more of, brands, categories and networks.  The method and system uses a Principal Component Analysis (PCA) model for computing the performance indicator.

Description

Disclosed is a method and system for tracking performance of a brand, a product category, or an online network.  The method and system provides a unified framework to combine multiple signals to estimate a performance indicator for one or more of, brands, categories and networks.  The method and system uses a Principal Component Analysis (PCA) model for computing the performance indicator.

In accordance with the method and system, multiple signals associated with a brand, a product category or a network associated with the brand is analyzed.  The multiple signals may include one or more signal resources, including but not limited to, search volume, content volume, social volume and content consumption volume.  Each of the signal resources may be classified for each user segment and may include one or more of, but not limited to, age, gender and device.

It is assumed that N signals are available in a T time period and represented as , wherein  is a vector of length  containing  signals associated with the one of, the brand, the category and the network.  Thereafter, a single indicator that represents a general trend in  is estimated. The single indicator  is modeled as a latent factor that controls signals  using the following equation:

wherein  is a vector of length  and  is a noise vector.

With proper assumptions[1], the single indicator  is estimated as a mean of principal components.  Covariance matrix of is calculated as:

wherein summation is from  to  and  is an average of .  Additionally, each element is a mean corresponding to one of, a brand, a category and a network signal. 

This is followed by calculation of first eigenvector of  that is represented by u.  The  is used to calculate  as,

The single indicator  is a linear combination of all signals  and the combination of weights is represented by vector .  

The variation of the single indicator  can be further controlled by fitting an Auto-Regressive Integrated Moving Average (ARIMA) model.  This ensures smoothening of the single indicator .  For example, an ARIMA (0, 1, 2) model is fitted with the single indicator  as follows:

wherein w[t] is white noise sequence and smoothened  is one step forecast of the fitted ARIMA (0, 1, 2) model.

In an embodiment, a ratio of brand indicator and category indicator may be...