1. Founder’s Technical Career
Since early 2012, we have been engaged in algorithm development related to Google and Bing. In 2014, we ventured into the e-commerce and media platform data management sector. Over the course of five years, from 2014 to 2019, we underwent a transformational journey from algorithm development and platform operation to business data training, ultimately achieving independent commercialization. During this period, we successfully developed and launched the first algorithmic system specifically designed for e-commerce data analysis, officially entering commercial markets.
2. Commercialization
In 2019, our operations officially commenced in Shenzhen, China.
2.1 First Team
The initial technical team comprised algorithm engineers and professionals in computational linguistics, including three algorithm developers and two computational linguistics graduate students.
2.2 Data Resources
Our original training data originates from diverse sources such as Wikipedia, Amazon, Google, Bing, and historical records from prominent global brands. These datasets primarily focus on product information and are complemented by authoritative language corpora like the BBC corpus, the BNC, the ANC, and contributions from renowned academic corpus labs. This ensures the standardization and professionalism of our training data across multiple language domains and specialized fields, aligning with commercial cooperation requirements and effectively utilized throughout our technical development process.
2.3 Current Technical Status
We maintain an autonomous and comprehensive product information corpus across all categories, with training completed. Additionally, we have developed specialized algorithmic models for advertising bid calculation across platforms like Amazon, Walmart, and Google, encompassing over 30 bidding matrix computation models. Furthermore, we possess a technical keyword matching library, utilized to counter keyword matching recognition algorithms, and maintain a transactional database encompassing all brands within existing platforms, with data cleansing completed.
+ Product Information Corpus
Primarily intended to construct a larger digital information matrix, applicable for advertising and SEO retrieval optimization, it exceeds 2.66 billion data nodes sourced from various languages.
Advertising Bidding Calculation Models (CPC)
Utilized for constructing automatic advertising bidding matrices, facilitating the acquisition of traffic and amplifying traffic matching within a fixed bidding range.
+Technical Keyword Matching Library
Designed to enhance keyword matching accuracy while simultaneously improving keyword relevance, enabling lower bidding on major e-commerce platforms and thus reducing advertising costs for merchants. It encompasses common misspelled words, erroneous inputs, and specific combinations of singular and plural terms.
+Branding Database
Facilitates comprehensive capture of brand interaction behaviors and fortifies the market share of proprietary brands.
3. Business Basics
3.1 Customer Service
In Shenzhen Huizhou, China, we maintain self-contained data center facilities solely designated for algorithm development and training, operating independently from the internet. We work with five auxiliary developers, two technical managers, and six technical operators for general data management and update upload.
3.2 Technical Support
Our team of five natural language processing development engineers in Seattle is responsible for algorithm development and updates, as well as data interaction.