Computer artificial intelligence information filtering system
By using a computer-based artificial intelligence information filtering system and optimizing through deep learning and user feedback, the problems of low accuracy and privacy security in information filtering systems have been solved. This results in efficient, personalized, and fair information filtering that adapts to information changes and user needs.
Patent Information
- Application Number
- CN202511751541.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-26
- Publication Date
- 2026-02-24
AI Technical Summary
Existing information filtering systems suffer from low filtering accuracy, incompatibility with user needs, algorithmic limitations, and user privacy and security issues when faced with massive and ever-changing information, making it difficult to achieve efficient and accurate information filtering.
The system employs a computer-based artificial intelligence information filtering system, which includes modules for data collection, preprocessing, information filtering, user feedback, and system optimization. It utilizes deep learning algorithms, convolutional neural networks, recurrent neural networks, and clustering algorithms, combined with user feedback and system optimization, to achieve personalized information filtering and privacy protection.
It improves the accuracy and efficiency of information filtering, provides personalized information customization, protects user privacy, ensures the fairness and diversity of information, and has good scalability and sustainability.
Smart Images
Figure BDA0005707070930000051 
Figure BDA0005707070930000052 
Figure BDA0005707070930000053
Abstract
Description
Technical Field
[0001] This invention relates to the field of artificial intelligence technology, specifically to a computer artificial intelligence information filtering system. Background Technology
[0002] With the advent of the information age, we are bombarded with all kinds of information every day. Whether it's the internet, social media, or email, people need to process a massive amount of information daily. While this information provides many conveniences for our work and lives, it also brings an important question: how to effectively filter and manage this vast amount of information.
[0003] Traditional information filtering systems primarily rely on predefined rules, keywords, or group feedback. While these methods can address the issue to some extent, filtering information based on predefined rules has significant limitations, such as low filtering accuracy and a mismatch between the filtering effect and the user's actual needs. This is especially true in the internet environment, where information is vast and updated rapidly, where traditional information filtering methods face even greater challenges.
[0004] In recent years, with the development of artificial intelligence technologies, machine learning methods such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), and Long Short-Term Memory Networks (LSTM), as well as clustering algorithms in unsupervised learning, have provided entirely new ideas. These methods, by automatically learning and recognizing data features, can achieve more precise personalized information filtering based on users' individual preferences.
[0005] However, even with these methods, overcoming the limitations of algorithmic technology and improving the quality, accuracy, and efficiency of information filtering remains a challenge in the face of massive and ever-changing information. Furthermore, ensuring the fairness, balance, and user privacy of filtering algorithms are also issues that need to be considered.
[0006] Therefore, there is an urgent need for a new type of computer information filtering system based on advanced artificial intelligence technology that can achieve efficient and accurate information filtering while taking into account fairness, balance, and user privacy. Summary of the Invention
[0007] The purpose of this invention is to provide a computer artificial intelligence information filtering system to solve the problems mentioned in the background art.
[0008] To achieve the above objectives, the present invention provides the following technical solution: a computer artificial intelligence information filtering system, comprising a data collection module, a data preprocessing module, an information filtering module, a user feedback module, and a system optimization module;
[0009] The data collection module is responsible for collecting data from various sources, including the Internet and databases. It crawls designated websites, social media, and news sites according to specific crawling strategies to obtain the latest information and data.
[0010] The data preprocessing module performs preprocessing operations such as cleaning and format conversion on the collected raw data, removes useless HTML tags, verifies and repairs data integrity, and unifies different data formats into a data format that can be processed internally by the system.
[0011] The information filtering module uses artificial intelligence technology, such as deep learning algorithms, to learn and understand information, and then filters the information according to preset rules and logic.
[0012] The user feedback module is responsible for collecting and processing user feedback on the information filtering results; users can evaluate the filtering results and mark information that is filtered incorrectly.
[0013] The system optimization module is responsible for analyzing user feedback and system operation logs, identifying problems and areas for improvement, and then adjusting and optimizing the system.
[0014] Preferably, the data collection module includes:
[0015] The source configuration component is responsible for defining and managing the configuration information of the data source, allowing users to input or modify the relevant parameters of the websites, API interfaces, databases and other information sources to be crawled. The source configuration component enables the data collection module to flexibly connect to a variety of data sources.
[0016] The scheduler component is responsible for arranging the execution of data collection tasks according to pre-set rules and schedules. The scheduler component ensures the orderly execution of data collection operations and provides a retry mechanism when a task fails.
[0017] The web crawler engine component is the core component responsible for executing data crawling tasks. It can read the parameters provided by the source configuration component, access the specified website or API, and then parse the page content or response data according to the predefined rules to extract the required information.
[0018] The data parser component is responsible for converting the raw data extracted by the crawler engine into a structured format.
[0019] Preferably, the data collection module further includes:
[0020] Once the data is successfully parsed into a structured format, the storage interface component is responsible for saving it to the specified storage system. The storage interface component can also provide caching functionality to reduce the access pressure on the data source and improve data access speed.
[0021] The error handling and logging component is responsible for capturing these exceptions, recording detailed log information, and taking appropriate countermeasures based on the configuration.
[0022] Preferably, the data preprocessing module includes:
[0023] The data cleaning component is used to remove noise, outliers, and irrelevant information from the data. The data cleaning component uses regular expressions to identify and delete unnecessary characters and data segments.
[0024] The data validation component is responsible for checking whether the collected data meets the expected specifications. If the data does not meet the validation rules, it will be marked as abnormal and the data repair process will be triggered.
[0025] The data repair component repairs data that has been found to be problematic after verification.
[0026] The data conversion component is responsible for converting raw data into a format suitable for a specific application.
[0027] The data integration component is responsible for combining and integrating this data to form a complete and consistent view;
[0028] The API interface and UI interface provide the means to operate the data preprocessing module, making it easy for users to execute and manage data preprocessing tasks.
[0029] Preferably, the information filtering module includes:
[0030] The rule engine is the core of the information filtering module. It determines whether data meets specific conditions based on predefined rules, and can quickly classify and filter data.
[0031] Keyword filtering component, used for filtering text data, identifies and filters data containing a set of keywords or phrases;
[0032] Pattern recognition components use machine learning or statistical models to identify specific patterns or anomalies in data;
[0033] The content analytics component is responsible for in-depth analysis of data content to extract valuable information, including sentiment analysis, topic modeling, entity recognition, and image and video analysis techniques.
[0034] To meet the personalized needs of different users, information filtering modules typically offer user-defined filtering functionality, allowing users to set filtering rules or adjust existing filtering rules according to their own requirements.
[0035] Preferably, the rules of the rule engine are:
[0036] Content-based rules, including keyword matching and regular expressions;
[0037] Structure-based rules, including data type and field integrity;
[0038] Time-based rules, including data update time;
[0039] The rule engine can assign a weight to each rule. During filtering, the data is scored based on the weight of the rule. Data with higher scores is considered to be more suitable for filtering conditions, making the rule engine more flexible and better able to adapt to different use cases.
[0040] The following formula is used to calculate the weights:
[0041] Score=Σ(WjI(Rj))
[0042] Where Wj is the weight of the j-th rule, Rj represents the j-th rule, and I(Rj) is an indicator function. If the data meets the j-th rule, then I(Rj) = 1, otherwise I(Rj) = 0. In this way, the score of each data is the sum of the weights of all the rules that are satisfied.
[0043] Preferably, the keyword filtering component is used to identify and filter text data containing specific keywords or phrases, including TF-IDF-based weight settings. By setting the weight of keywords through TF-IDF, it is possible not only to identify data containing keywords, but also to measure the importance of keywords in the data. The formula is: `TF-IDF = TFIDF` where TF represents the frequency of keyword occurrence in the text, and IDF reflects the distinctiveness of the keyword.
[0044] Preferably, the pattern recognition component is used to extract specific patterns or rules from data, including:
[0045] Image recognition based on convolutional neural networks:
[0046] It employs a combined structure of multiple convolutional layers, pooling layers, and fully connected layers;
[0047] The formula for convolution is:
[0048]
[0049] Where I represents the input image, K represents the convolution kernel, and (i,j) are the image coordinates; Time series pattern recognition based on recurrent neural networks and long short-term memory networks: LSTM units are used to process time series data, which can capture long-term dependencies; The LSTM formula includes a forget gate, an input gate, candidate memory units, and an output gate:
[0050] f t =σ(W f ·[h t-1 x t ]+b f )
[0051] i t =σ(W i ·[h t-1 ,x t ]+b i )
[0052]
[0053] o t =σ(W o ·[h t-1 ,x t ]+b o )
[0054] h t =o t ·tanh(C t )
[0055] Unsupervised learning pattern recognition based on clustering algorithms:
[0056] Clustering algorithms such as K-means or DBSCAN are used to discover potential population patterns in the data;
[0057] The objective function for K-means clustering is:
[0058]
[0059] Where k is the number of clusters, Ci is the i-th cluster, and ui is the center of the i-th cluster.
[0060] A novel computer-based artificial intelligence information filtering system, based on the latest artificial intelligence technology, achieves the following beneficial effects:
[0061] 1. Improve the accuracy and efficiency of information filtering: By employing advanced machine learning algorithms, such as deep learning networks (including convolutional neural networks, recurrent neural networks, etc.), this system can more accurately understand and analyze the content and structure of different types of information, such as text, images, and videos. This not only improves the accuracy of information filtering, but also, due to the algorithm's self-learning and adaptive capabilities, it can continuously optimize the process to adapt to new information types and user preferences, thereby improving the efficiency of information processing.
[0062] 2. Personalized Information Customization: Utilizing advanced data analysis technology, this system can personalize and recommend information based on users' historical behavior data and preferences. This means that users will receive more customized information that matches their personal interests, greatly enhancing the user experience and saving users time in filtering information.
[0063] 3. Protecting user privacy: In designing this system, special attention is paid to user privacy and security; by adopting advanced data encryption and anonymization technologies, the security and privacy of user data are ensured, so that users can enjoy personalized information services without worrying about the risk of personal data leakage.
[0064] 4. Fairness and balance: By introducing diversity and fairness indicators, this system ensures that information filtering does not lead to information bubbles or exacerbate biases; it can provide users with a wide range of diverse information perspectives, helping them obtain more balanced and comprehensive information, thereby making more rational decisions.
[0065] 5. Easy to expand and update: Built on a sustainable and scalable architecture, this system is easy to integrate new algorithms and technologies to meet future changes in information filtering needs; this means that the system has strong future compatibility and can continue to evolve with technological advancements and market changes.
[0066] In summary, the computer artificial intelligence information filtering system disclosed in this invention not only improves the accuracy and efficiency of information filtering, but also enhances user experience, protects user privacy, ensures the fairness and diversity of information, and has good scalability and sustainability, providing users with a safe, efficient, and personalized information consumption environment. Detailed Implementation
[0067] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0068] This invention provides a technical solution: a computer artificial intelligence information filtering system, comprising a data collection module, a data preprocessing module, an information filtering module, a user feedback module, and a system optimization module;
[0069] The data collection module is responsible for collecting data from various sources, including the internet and databases. Based on specific crawling strategies, it crawls designated websites, social media, and news sites to obtain the latest information and data.
[0070] The data collection module includes: a source configuration component, a scheduler component, a crawler engine component, a data parser component, a storage interface component, and an error handling and logging component. The source configuration component is responsible for defining and managing the configuration information of the data source, allowing users to input or modify relevant parameters of the websites, API interfaces, databases, and other information sources to be crawled. This component enables the data collection module to flexibly connect to a variety of data sources. The scheduler component is responsible for scheduling the execution of crawling tasks according to pre-set rules and schedules. The scheduler component ensures the orderly execution of data collection operations and provides a retry mechanism when tasks fail. The crawler engine component is responsible for executing data... The core components of the crawling task can read parameters provided by the source configuration component, access the specified website or API, and then parse the page content or response data according to predefined rules to extract the required information. The data parser component is responsible for converting the raw data extracted by the crawler engine into a structured format. Once the data is successfully parsed into a structured format, the storage interface component is responsible for saving it to the specified storage system. The storage interface component can also provide caching functionality to reduce the access pressure on the data source and improve data access speed. The error handling and logging component is responsible for capturing these exceptions, recording detailed log information, and taking corresponding countermeasures according to the configuration.
[0071] The data preprocessing module performs preprocessing operations such as cleaning and format conversion on the collected raw data, removes useless HTML tags, verifies and repairs data integrity, and unifies different data formats into a data format that can be processed internally by the system.
[0072] The data preprocessing module includes: a data cleaning component, a data validation component, a data repair component, a data transformation component, a data integration component, and an API interface / UI interface. The data cleaning component removes noise, outliers, and irrelevant information from the data, using methods such as regular expressions to identify and delete unnecessary characters and data segments. The data validation component checks whether the collected data conforms to expected specifications; if data does not meet validation rules, it is marked as abnormal, triggering the data repair process. The data repair component repairs data found to have problems after validation. The data transformation component converts the raw data into a format suitable for a specific application. The data integration component combines and integrates these data to form a complete and consistent view. The API interface / UI interface provides the means to operate the data preprocessing module, allowing users to easily execute and manage data preprocessing tasks.
[0073] The information filtering module uses artificial intelligence technology, such as deep learning algorithms, to learn and understand information, and then filters the information according to preset rules and logic.
[0074] The information filtering module includes: a rule engine, a keyword filtering component, a pattern recognition component, a content analysis component, and a user-defined filtering component. The rule engine is the core of the information filtering module; it determines whether data meets specific conditions based on predefined rules, enabling rapid data classification and filtering. The keyword filtering component filters text data by setting a set of keywords or phrases to identify and filter data containing those keywords. The pattern recognition component uses machine learning or statistical models to identify specific patterns or anomalies in the data. The content analysis component is responsible for in-depth analysis of the data content to extract valuable information, including sentiment analysis, topic modeling, entity recognition, and image and video analysis techniques. To meet the personalized needs of different users, the information filtering module typically provides user-defined filtering functionality, allowing users to set filtering rules according to their own needs or adjust existing filtering rules.
[0075] The rules of the rule engine are:
[0076] Content-based rules, including keyword matching and regular expressions;
[0077] Structure-based rules, including data type and field integrity;
[0078] Time-based rules, including data update time;
[0079] The rule engine can assign a weight to each rule. During filtering, the data is scored based on the weight of the rule. Data with higher scores is considered to be more suitable for filtering conditions, making the rule engine more flexible and better able to adapt to different use cases.
[0080] The following formula is used to calculate the weights:
[0081] Score=Σ(WjI(Rj))
[0082] Where Wj is the weight of the j-th rule, Rj represents the j-th rule, and I(Rj) is an indicator function. If the data meets the j-th rule, then I(Rj) = 1, otherwise I(Rj) = 0. In this way, the score of each data is the sum of the weights of all the rules that are satisfied.
[0083] The keyword filtering component is used to identify and filter text data containing specific keywords or phrases, including TF-IDF-based weight settings. By setting the weight of keywords through TF-IDF, it is possible not only to identify data containing keywords, but also to measure the importance of keywords in the data. The formula is: `TF-IDF = TFIDF` where TF represents the frequency of keyword occurrence in the text, and IDF reflects the distinctiveness of the keyword.
[0084] The pattern recognition component is used to extract specific patterns or rules from data, including:
[0085] Image recognition based on convolutional neural networks:
[0086] It employs a combined structure of multiple convolutional layers, pooling layers, and fully connected layers;
[0087] The formula for convolution is:
[0088]
[0089] Where I represents the input image, K represents the convolution kernel, and (i,j) are the image coordinates;
[0090] Time series pattern recognition based on recurrent neural networks and long short-term memory networks:
[0091] Using LSTM units to process time series data can capture long-term dependencies.
[0092] The formula for LSTM includes a forget gate, an input gate, candidate memory units, and an output gate:
[0093] f t =σ(W f ·[h t-1 x t ]+b f )
[0094] i t =σ(W i ·[h t-1 ,x t ]+b i )
[0095]
[0096] o t =σ(W o ·[h t-1 x t ]+b o )
[0097] h t =o t ·tanh(C t )
[0098] Unsupervised learning pattern recognition based on clustering algorithms:
[0099] Clustering algorithms such as K-means or DBSCAN are used to discover potential population patterns in the data;
[0100] The objective function for K-means clustering is:
[0101]
[0102] Where k is the number of clusters, Ci is the i-th cluster, and ui is the center of the i-th cluster.
[0103] The user feedback module is responsible for collecting and processing user feedback on the information filtering results; users can evaluate the filtering results and mark information that is filtered incorrectly.
[0104] The system optimization module is responsible for analyzing user feedback and system operation logs, identifying problems and areas for improvement, and then adjusting and optimizing the system.
[0105] In practical applications, the aim is to improve employee efficiency and reduce information overload by filtering out important information relevant to the business.
[0106] Experimental setup
[0107] Overall dataset: 100,000 messages taken from the company's internal communications, including emails, messages, and reports.
[0108] Target Classification: For our system, the goal is to filter out all information related to Project X, which is a current high-priority task for the company.
[0109] Training and testing allocation: 80% of the data is randomly selected from the overall dataset for training, and the remaining 20% is used to test the system's performance.
[0110] Performance metrics:
[0111] Accuracy: The proportion of correctly classified information out of the total number of information items.
[0112] Precision: The proportion of items correctly identified as related to item X out of the total number of items identified as related to item X.
[0113] Recall: The proportion of items correctly identified as related to item X out of the total number of items related to item X.
[0114] Example Results
[0115] 1. Data preparation: 100,000 text messages were extracted from the company's communication records, and 2,000 of them were manually labeled as related to Project X based on keywords, people involved in the aforementioned projects, and timelines.
[0116] 2. Training the AI model: A deep learning model is used, trained with word vectors as feature input. The model's performance on the training set gradually improves until the accuracy exceeds 95%.
[0117] 3. Testing and Evaluation:
[0118] On the independent test set, the system achieved an accuracy of 92%, meaning it can efficiently identify the vast majority of relevant and irrelevant information.
[0119] The accuracy rate is 89%, which means that 89% of the information identified by the system is accurate.
[0120] A recall rate of 90% means that the system can retrieve 90% of all truly relevant information.
[0121] 4. Optimization and Adjustment: Based on the test results, the parameters were further fine-tuned, and the diversity and quantity of training data were increased in order to further improve system performance.
[0122] Conclusion: The provided data demonstrates that the described "computer artificial intelligence information filtering system" can effectively process large-scale information, improving the efficiency and accuracy of information processing by filtering data relevant to specific projects. Furthermore, through continuous training and optimization, the system can continuously improve its predictive performance, showcasing significant application potential.
[0123] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A computer artificial intelligence information filtering system, characterized in that, It includes a data collection module, a data preprocessing module, an information filtering module, a user feedback module, and a system optimization module; The data collection module is responsible for collecting data from various sources, including the Internet and databases. It crawls designated websites, social media, and news sites according to specific crawling strategies to obtain the latest information and data. The data preprocessing module performs preprocessing operations such as cleaning and format conversion on the collected raw data, removes useless HTML tags, verifies and repairs data integrity, and unifies different data formats into a data format that can be processed internally by the system. The information filtering module uses artificial intelligence technology, such as deep learning algorithms, to learn and understand information, and then filters the information according to preset rules and logic. The user feedback module is responsible for collecting and processing user feedback on the information filtering results; users can evaluate the filtering results and mark information that is filtered incorrectly. The system optimization module is responsible for analyzing user feedback and system operation logs, identifying problems and areas for improvement, and then adjusting and optimizing the system.
2. The computer artificial intelligence information filtering system according to claim 1, characterized in that, The data collection module includes: The source configuration component is responsible for defining and managing the configuration information of the data source, allowing users to input or modify the relevant parameters of the websites, API interfaces, databases and other information sources to be crawled. The source configuration component enables the data collection module to flexibly connect to a variety of data sources. The scheduler component is responsible for arranging the execution of data collection tasks according to pre-set rules and schedules. The scheduler component ensures the orderly execution of data collection operations and provides a retry mechanism when a task fails. The web crawler engine component is the core component responsible for executing data crawling tasks. It can read the parameters provided by the source configuration component, access the specified website or API, and then parse the page content or response data according to the predefined rules to extract the required information. The data parser component is responsible for converting the raw data extracted by the crawler engine into a structured format.
3. The computer artificial intelligence information filtering system according to claim 2, characterized in that, The data collection module also includes: Once the data is successfully parsed into a structured format, the storage interface component is responsible for saving it to the specified storage system. The storage interface component can also provide caching functionality to reduce the access pressure on the data source and improve data access speed. The error handling and logging component is responsible for capturing these exceptions, recording detailed log information, and taking appropriate countermeasures based on the configuration.
4. The computer artificial intelligence information filtering system according to claim 3, characterized in that, The data preprocessing module includes: The data cleaning component is used to remove noise, outliers, and irrelevant information from the data. The data cleaning component uses regular expressions to identify and delete unnecessary characters and data segments. The data validation component is responsible for checking whether the collected data meets the expected specifications. If the data does not meet the validation rules, it will be marked as abnormal and the data repair process will be triggered. The data repair component repairs data that has been found to be problematic after verification. The data conversion component is responsible for converting raw data into a format suitable for a specific application. The data integration component is responsible for combining and integrating this data to form a complete and consistent view; The API interface and UI interface provide the means to operate the data preprocessing module, making it easy for users to execute and manage data preprocessing tasks.
5. A computer artificial intelligence information filtering system according to claim 4, characterized in that, The information filtering module includes: The rule engine is the core of the information filtering module. It determines whether data meets specific conditions based on predefined rules, and can quickly classify and filter data. Keyword filtering component, used for filtering text data, identifies and filters data containing a set of keywords or phrases; Pattern recognition components use machine learning or statistical models to identify specific patterns or anomalies in data; The content analytics component is responsible for in-depth analysis of data content to extract valuable information, including sentiment analysis, topic modeling, entity recognition, and image and video analysis techniques. To meet the personalized needs of different users, information filtering modules typically offer user-defined filtering functionality, allowing users to set filtering rules or adjust existing filtering rules according to their own requirements.
6. A computer artificial intelligence information filtering system according to claim 5, characterized in that, The rules of the rule engine are: Content-based rules, including keyword matching and regular expressions; Structure-based rules, including data type and field integrity; Time-based rules, including data update time; The rule engine can assign a weight to each rule. During filtering, the data is scored based on the weight of the rule. Data with higher scores is considered to be more suitable for filtering conditions, making the rule engine more flexible and better able to adapt to different use cases. The following formula is used to calculate the weights: Score=Σ(WjI(Rj)) Where Wj is the weight of the j-th rule, Rj represents the j-th rule, and I(Rj) is an indicator function. If the data meets the j-th rule, then I(Rj) = 1, otherwise I(Rj) = 0. In this way, the score of each data is the sum of the weights of all the rules that are satisfied.
7. A computer artificial intelligence information filtering system according to claim 6, characterized in that, The keyword filtering component is used to identify and filter text data containing specific keywords or phrases, including TF-IDF-based weight settings. By setting the weight of keywords through TF-IDF, it is possible not only to identify data containing keywords, but also to measure the importance of keywords in the data. The formula is: `TF-IDF = TFIDF` where TF represents the frequency of keyword occurrence in the text, and IDF reflects the distinctiveness of the keyword.
8. A computer artificial intelligence information filtering system according to claim 7, characterized in that, The pattern recognition component is used to extract specific patterns or rules from data, including: Image recognition based on convolutional neural networks: It employs a combined structure of multiple convolutional layers, pooling layers, and fully connected layers; The formula for convolution is: Where I represents the input image, K represents the convolution kernel, and (i,j) are the image coordinates; Time series pattern recognition based on recurrent neural networks and long short-term memory networks: Using LSTM units to process time series data can capture long-term dependencies. The formula for LSTM includes a forget gate, an input gate, candidate memory units, and an output gate: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) the t =σ(W o ·[h t-1 ,x t ]+b o ) h t = no t ·fish(C) t ) Unsupervised learning pattern recognition based on clustering algorithms: Clustering algorithms such as K-means or DBSCAN are used to discover potential population patterns in the data; The objective function for K-means clustering is: Where k is the number of clusters, Ci is the i-th cluster, and ui is the center of the i-th cluster.