WO2023029189A1 - 面向dikw内容的意图驱动交互填表方法 - Google Patents

面向dikw内容的意图驱动交互填表方法 Download PDF

Info

Publication number
WO2023029189A1
WO2023029189A1 PCT/CN2021/127300 CN2021127300W WO2023029189A1 WO 2023029189 A1 WO2023029189 A1 WO 2023029189A1 CN 2021127300 W CN2021127300 W CN 2021127300W WO 2023029189 A1 WO2023029189 A1 WO 2023029189A1
Authority
WO
WIPO (PCT)
Prior art keywords
filling
dikp
model
information
intention
Prior art date
Application number
PCT/CN2021/127300
Other languages
English (en)
French (fr)
Inventor
段玉聪
黄越
Original Assignee
海南大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 海南大学 filed Critical 海南大学
Priority to US17/564,266 priority Critical patent/US20230065902A1/en
Publication of WO2023029189A1 publication Critical patent/WO2023029189A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/10—Text processing
    • G06F40/166—Editing, e.g. inserting or deleting
    • G06F40/174—Form filling; Merging
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367—Ontology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Definitions

  • the invention relates to the technical field of information processing, in particular to an intent-driven interactive form filling method for DIKW content.
  • the sorted data is matched and filled in with key values, and the data items in the Excel table are automatically filled into a single Excel/word form one by one.
  • a data set contract is automatically generated in batches, which does not reflect the intelligence it should have on how to fill in the content.
  • the intention of the person filling the form is not considered during the filling process, and fillings that are contrary to the wishes of the person filling the form may occur Behavior, likewise, does not judge the intention of analyzing the form, and does not consider whether it may bring adverse consequences to the person who fills the form after filling it.
  • the present invention proposes an intention-driven interactive form-filling method oriented to DIKW content, fully considers the intentions of the form-filler and the form, and solves various problems generated during the form-filling process with intention-driven, realizes intelligence and improves rationality .
  • the intent-driven interactive form filling method for DIKW content includes the following steps:
  • Step S1 constructing the first DIKP model according to the information of the person filling the form, and constructing the second DIKP model according to the form information;
  • Step S2 perform intent comparison and value judgment according to the first DIKP model and the second DIKP model, and obtain the filling value level of the table;
  • Step S4 performing feedback verification on the completed form information.
  • the construction process of the first DIKP model in step S1 is: judging whether there is a previous DIKP model of the person filling the form in the system, if not, then establishing the first DIKP according to the data and intention currently input by the person filling the form model; if there is a previous DIKP model of the person filling the form in the system, the current input data and intent are used to supplement and update the previous DIKP model, and obtain the first DIKP model.
  • the second DIKP model in the step S1 obtains the intention of the whole table and each table item according to the table category, internal table items, positional relationship, topological structure and alternative items, and constructs the second DIKP model accordingly.
  • the specific steps of the step S2 are: comparing the intention of the form with the intention of the person filling the form, comparing the semantic analysis with the DIKP model to determine the difference between the intentions, and judging whether the filling of the form generates income according to the intention of the person filling the form, according to The size of the difference and whether there is a difference to judge the filling value level of the table.
  • the filling value level includes three levels: high, general, and low, and the difference between the intention of the form and the intention of the form filler is small, and it can be judged according to the intention of the form filler to bring benefits to the form filler, that is, The level of filling value is high; the difference between the intention of the form and the intention of the person filling the form is small, and it is difficult to judge whether it will bring benefits according to the intention of the person filling the form, that is, the level of filling value and general; the intention of the form and the intention of the person filling the form If the difference is large, the filling value level is low.
  • the method of reducing certainty is used to fill the table information, and the specific steps include:
  • Step S311 obtaining the first data about the person filling the form in the first DIKP model
  • Step S312 obtaining a plurality of entries associated with the first data in the table in the second DIKP model
  • Step S313 traversing the first DIKP model for the plurality of associated entries in step S312, and finding information associated with the first data according to the mapping relationship;
  • Step S314 filling the first data into the corresponding entry according to the information associated with the first data.
  • the fuzzy transfer method is used to fill the form information, and the specific steps are: obtaining the second data about the person who filled the form in the first DIKP model, and using the cutting data, information, The inclusion, cascade, partial order relationship between knowledge, or the way of fuzzy change is filled into the table.
  • the fuzzy modification includes converting deterministic type into probability type, independent type into comparative type, overall type into local type and numerical type into range type.
  • a defensive filling method is used to fill the form information, and the defensive filling method includes not filling and filling the form by fuzzy transfer.
  • Step S41 traversing the first DIKP model and the second DIKP model, and judging whether the filled content meets the intention of the person filling the form;
  • Step S42 when the filled content does not meet the intention of the person filling the form, modify the content that does not meet the intention by means of intention coverage or blurred modification.
  • the present invention provides an intent-driven interactive form filling method oriented to DIKW content, and constructs a DIKP model for the form filler and the form, wherein the DIKP model includes the data, information, knowledge and intention of the form filler and the form, By comparing the intentions of the two DIKP models and making value judgments, the filling value level of the table can be obtained. According to the different filling value levels, the table information can be filled by reducing certainty, fuzzy transmission and defensive filling. Finally, the filled information Carry out feedback verification, by combining the DIKP system into the automatic form filling, fully consider the intention of the form filler and the form, and realize the intelligence and rationalization of form content filling.
  • FIG. 1 is a flow chart of the intent-driven interactive form filling method oriented to DIKW content in the present invention.
  • the intent-driven interactive form filling method oriented to DIKW content includes the following steps:
  • Step S1 constructing the first DIKP model according to the information of the person who filled the form, and constructing the second DIKP model according to the information of the form.
  • DIKP For the DIKP model, it includes data system, information system, knowledge system, and intention system.
  • the data in the data system are discrete elements that are directly implemented, and do not reflect specific meaning without context.
  • the combination of intentions will generate information; the information in the information system is composed of data and intentions, which is a response to data according to specific intentions, and is a directional expression. Multiple intentions can be related to data or information.
  • Concept transformation from data type to information type is achieved by associating target data with at least one intent; intent is an implicit or explicit purpose or goal that humans have related to specific things.
  • intent Mainly associated with data, single data or multiple data can be combined with one or more intents. The main relationships between intentions themselves are AND or NOT and inclusion.
  • An intention can be divided into several sub-intentions; knowledge is obtained from data and information through structured and formalized statistics and derivation, and further condensed on the basis of information to form A kind of knowledge rule, which reflects a certain regularity, has a certain stability and reusability, and data, information, and knowledge can be combined and transformed with each other.
  • the construction process of the first DIKP model in step S1 is: judging whether there is a previous DIKP model of the person filling the form in the system, if not, then establishing the first DIKP according to the data and intention currently input by the person filling the form model; if there is a previous DIKP model of the person who filled the form in the system, then use the currently input data and intent to supplement and update the previous DIKP model, and obtain the first DIKP model, the second DIKP model in the step S1 according to the form
  • the categories, internal table items, positional relationships, topological structures, and alternatives obtain the intent of the entire table and each table item, and construct the second DIKP model based on this.
  • the final system is based on a map Stored in a way to obtain the first DIKP model
  • the construction process of the second DIKP model is similar to the construction process of the first DIKP model, after the construction of the second DIKP system, it can be used to analyze the difference between the intention of the person and the person who filled the form, and Whether it will have an adverse impact on the person filling the form is used to make a value judgment.
  • the general internal framework of the DIKP system is the same.
  • the number of edges passed from the initial node to the target node is called distance, and the number of passed nodes is called depth; 2) A structure composed of some nodes and edges is called a structure; 3) The edge from a node is called an outgoing edge, also called a branch, and its number is called the out-degree of the node, and the edge entering a node is called an incoming edge, and its number is called The in-degree of the node; 4) The distance and depth from the initial node to the target node represent the cognitive distance, and the out-degree sum of all nodes in the depth is called the deviation coefficient.
  • the cognitive distance is too long and the deviation coefficient is too high, it will cost to find the target node The cost will be too high, but the target accuracy will increase; 5) The out-degree of the node and its child nodes and the probability deviation factor of the node, the greater the probability deviation factor, the accuracy of the target node will decrease.
  • Step S2 perform intent comparison and value judgment according to the first DIKP model and the second DIKP model, and obtain the filling value level of the form.
  • the specific steps are: compare the intent of the form with the intent of the person filling the form, and compare and judge through the semantic analysis and the DIKP model At the same time, according to the intention of the person filling the form, it is judged whether the filling of the form generates income, and the filling value level of the form is judged according to the size of the difference and whether there is a difference.
  • the filling value level includes three levels: high, general, and low.
  • the difference between the intent and the person who fills in the form is small, and if it can be judged according to the person who filled the form that it will bring benefits to the person filling the form, it means that the level of filling value is high; the difference between the intention of the form and the person who filled the form is small, and it is difficult to If it is judged according to the intention of the person who fills out the form whether it will bring benefits, it means that the level of filling value is average; if the intention of the form is quite different from that of the person filling out the form, it means that the level of filling value is low.
  • Step S3 After obtaining the filling value level of the form, different methods are used to fill the form information according to the level of the filling value level.
  • the filling value level is high, the form information is filled with reduced certainty.
  • the fuzzy transfer method is used to fill the table information, and when the filling value level is low, the defensive filling method is used to fill the table information.
  • Step S311 obtaining the first data about the person filling the form in the first DIKP model
  • Step S312 obtaining a plurality of entries associated with the first data in the table in the second DIKP model
  • Step S313 traversing the first DIKP model for the plurality of associated entries in step S312, and finding information associated with the first data according to the mapping relationship;
  • Step S314 filling the first data into the corresponding entry according to the information associated with the first data.
  • Tom's data system has three data, DAT(18), DAT(5), and DAT(70). It is not clear what the three data mean when they are insufficient, incomplete and inaccurate.
  • items such as age and number can be filled in.
  • For the DAT (age) item traverse Tom’s first DIKP model, and find the DIKP content that may be related to age according to the mapping relationship.
  • the fuzzy transfer method is used to fill the form information, and the specific steps are: obtaining the second data about the person who filled the form in the first DIKP model, and using the cutting data, information, The inclusion, cascade, partial order relationship between knowledge, or the way of fuzzy change is filled into the table.
  • the person who fills the form may hope that his data information knowledge can only be obtained by the form to the necessary part, satisfy the intention, reduce the risk of leakage, and even do not want to Let the form know some complete data or information, and want to verify the authenticity and intent of the form.
  • the flow of data or information from the form filler to the form can be vaguely transferred, combined with the form filler's first DIKP model , to make certain changes to data, information or knowledge, to increase the probability deviation factor, or to cut off the inclusion, cascade or partial order relationship between data, information or knowledge.
  • Fuzzy transfer can delete the key nodes of the filled content during the filling process, destroying the past relationship of data, information or knowledge transfer.
  • the present invention also provides a way of fuzzy modification to fill in the table information. Fuzzy modification includes converting deterministic into probability type, convert independent type to comparative type, convert global type to partial type, and convert numeric type to range type.
  • the conversion from deterministic to probabilistic is to express a definite data or information in the form of probability, such as INF (Tom was fired by the previous unit) can be converted into INF (Tom may resign voluntarily or be fired In this case, Tom’s real resignation information cannot be completely determined when filling in the DAT (work experience) item, and the intention of the person filling the form is PUP (Tom does not want people to know that he is fired), and at the same time It also provides reference information to the form, which satisfies the intent of the form to a certain extent and meets the filling requirements. Probabilistic data requires the form to reason and judge the content based on the data, information or knowledge filled by the form filler and combined with its own DIKP system, and at the same time reduces The risk of data or information leakage of the person filling the form is reduced.
  • the transformation from independent type to comparative type is a partial order transformation, which transforms an independently expressed data or information into a form of comparison with other data or information, such as INF (Tom's height is 170cm), which is an independent information, It can be transformed into information in the comparative form of INF (Tom is 5cm taller than Jerry).
  • INF Tom's height is 170cm
  • INF Tom is 5cm taller than Jerry
  • the conversion from integral type to partial type is a process of de-integrity, which means that a certain data or information of the person filling the form is partially segmented, and the whole content is not provided to the other party, but a part of it is provided.
  • a certain data or information of the person filling the form is partially segmented, and the whole content is not provided to the other party, but a part of it is provided.
  • Tom's age is DAT(29)
  • Tom is a woman, and she doesn't want people to know that she is going to be 30 years old. It can be changed to DAT(2*), and the other part is hidden. Facing some form intentions such as PUP( Know whether the person filling the form is an adult), then the data DAT(2*) can meet the needs of the form and at the same time meet Tom's intention.
  • Numerical to range conversion is to hide some numerical data or information in a certain range, which can be hidden in different ranges according to the intention of the person filling the form.
  • Tom weight is 80kg, and his intention is not to let others know He is so heavy that he can convert the numerical data of DAT (80kg) into range data DAT (>60kg) to meet Tom's intention.
  • Tom is 16 years old this year and wants to enjoy some conveniences for minors, such as buying half-price tickets, etc., but he does not want to disclose accurate age information to strangers. Into the scope of minors, to meet Tom's intention.
  • a defensive filling method is used to fill the form information, and the defensive filling method includes not filling and filling the form by fuzzy transfer.
  • This step can also be called misleading filling.
  • This step can also be called misleading filling.
  • Filling measures When facing this kind of form, do not fill it directly when the filling value is extremely low, so as to avoid too much loss to the form filler.
  • step S4 Before the content is filled, it is necessary to verify the content of the data or information filled in by the person filling in the form. Since the data, information, intent and knowledge system have certain inclusion, cascading and partial order relationships, it is easy to check the content of a table item node. Filling violates the intention of another table item node, and the filled data or information is associated with other table items. Therefore, it is necessary to perform step S4 and perform feedback verification on the filled table information.
  • step S41 traversing the first DIKP model and the second DIKP model, judging whether the filled content conforms to the intention of the form filler;
  • Step S42 when the filled content does not conform to the intention of the form filler, modify the content that does not conform to the intention by means of intention coverage or fuzzy modification.
  • Intent coverage includes supplementary coverage mapping of the person filling the form according to the inclusion, cascade or partial order relationship. If the person who fills the form does not want people to know his age, the intention can be supplemented from DAT(age) by the cascade or partial order relationship of the data Mapped to DAT (birthday) and DAT (ID number), to avoid appearing in the filling of a certain table item against other intentions of the person filling the form; 2) Cut off the inclusion, cascade or partial order relationship, you can use the method in the fuzzy transfer Blur or change the data or information to be filled, cut off the relationship between the data or information to be filled, when the person who fills the form does not want people to know his age is DAT(29), when filling his date of birth, he can perform integral to partial conversion, Hide the year, such as DAT(5.7), or convert it in other ways so that it does not violate the intent of the person filling the form.
  • the present invention combines the DIKP system into the automatic form filling, and fills the form information according to the intentions of both the form filler and the form.
  • the intent of the form-filler so as to achieve intelligence and rationalization, reduce the tedious manual filling process of the form-filler, and protect the privacy of the form-filler.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种面向DIKW内容的意图驱动交互填表方法，包括以下步骤：根据填表人信息构建第一DIKP模型，根据表格信息构建第二DIKP模型；根据第一DIKP模型以及第二DIKP模型进行意图比较以及价值判断，获取表格的填充价值等级；根据填充价值等级的不同采用降低确定性、模糊传递以及防御填充方法进行表格信息填充；对填充完成的表格信息进行反馈核验，将DIKP体系结合到自动填表系统中，对信息进行自动化的填写，可以节省填表人手动填写信息的过程，实现填表的智能化，而在填写过程中充分考虑填表人的意图，使填写的信息不会违背填表人的意图，保护填表人的隐私。

Description

面向DIKW内容的意图驱动交互填表方法 技术领域
本发明涉及信息处理技术领域，特别涉及面向DIKW内容的意图驱动交互填表方法。
背景技术
随着当今世界的信息化和数据化发展，各种各样的表单层出不绝，吃饭点个外卖需要填写外卖表单，买高铁飞机票出游需要填写购票表单，各种app和网站的调查问卷，公司出差或请假等等，这些都可归为内容填充一类问题，当这些待填充内容比较多的时候，填表人填写时就会遇到重复填写、填写出错、嫌麻烦、泄露信息等常见问题，以表单为主的应用，最繁琐的工作就是需要填写很多表单字段，用户所填写的内容经常是重复的工作，并且有些字段操作了很多步骤只是为了获取同一个选项值。
现有自动填充技术大多都是限制在了数据或信息迁移的程度上，基本都是将整理好的数据按键值匹配填入，将Excel表中数据条逐条自动填写到单个Excel/word表单，录入一条数据成套合同批量自动生成，这对于如何填充内容并没有体现出应有的智能，此外，在填充过程中并没有对填表人的意图进行考虑，可能会产生与填表人意愿相反的填充行为，同样，也没有判断分析表格的意图，没有考虑填充之后是否可能给填表人带来不利后果。
每个人都具有属于自身的data(数据)、information(信息)、knowledge(知识)以及智慧(Wisdom)，代表着自身对外部客观事物的认知和理解，通过对数据、信息、知识以及智慧进行建模可以获得DIKW图谱体系，DIKW图谱是目前数据处理中较为新型的图谱模型，从DIKW图谱中结合个人的意图(purpose)可以获得DIKP图谱体系，DIKP图谱体系中的数据可用于存储、传输以及计算，然而目前并没有相关文献将DIKP图谱体系应用到智能填表中以解决填表繁琐、填充内容不正确、填充过程不考虑填表人意图等问题中。
发明内容
鉴以此，本发明提出面向DIKW内容的意图驱动交互填表方法，充分考虑填表人和表格的意图，以意图驱动来解决填表过程中产生的各类问题，实现智能化，提高合理性。
本发明的技术方案是这样实现的：
面向DIKW内容的意图驱动交互填表方法，包括以下步骤：
步骤S1、根据填表人信息构建第一DIKP模型，根据表格信息构建第二DIKP模型；
步骤S2、根据第一DIKP模型以及第二DIKP模型进行意图比较以及价值判断，获取表格的填充价值等级；
步骤S3、根据填充价值等级的不同采用降低确定性、模糊传递以及防御填充方法进行表格信息填充；
步骤S4、对填充完成的表格信息进行反馈核验。
优选的，所述步骤S1中的第一DIKP模型的构建过程为：判断系统中是否存在填表人以往的DIKP模型，若不存在，则根据填表人当前输入的数据和意图建立第一DIKP模型；若系统中存在填表人以往的DIKP模型，则利用当前输入的数据和意图对以往的DIKP模型进行补充更新，并获得第一DIKP模型。
优选的，所述步骤S1中的第二DIKP模型根据表格类别、内部表项、位置关系、拓扑结构以及备选项获取表格整体与各表项的意图，并以此构建第二DIKP模型。
优选的，所述步骤S2的具体步骤为：将表格意图以及填表人意图进行比较，通过语义分析与DIKP模型比较判断意图间的差异，同时根据填表人意图判断填充表格是否产生收益，根据差异大小以及是否产生差异来判断表格的填充价值等级。
优选的，所述填充价值等级包括高、一般、低三个等级，在表格的意图与填表人的意图差异较小，且能根据填表人意图判断给填表人带来收益的，即为填充价值等级高；表格的意图与填表人的意图差异较小，且难以根据填表人意 图判断是否带来收益的，即为填充价值等级与一般；表格的意图与填表人的意图差异较大的，即为填充价值等级低。
优选的，在判断填充价值等级为高时，采用降低确定性方法进行表格信息填充，具体步骤包括：
步骤S311、获取第一DIKP模型中关于填表人的第一数据；
步骤S312、获取第二DIKP模型中的表格与第一数据具有关联的多个表项；
步骤S313、针对步骤S312中具有关联的多个表项遍历第一DIKP模型，并根据映射关系找到与第一数据相关联的信息；
步骤S314、根据与第一数据相关联的信息将第一数据填充到对应的表项中。
优选的，在判断填充价值等级为一般时，采用模糊传递方法进行表格信息填充，具体步骤为：获取第一DIKP模型中关于填表人的第二数据，对第二数据采用切断数据、信息、知识之间的包含、级联、偏序关系方式或模糊更改的方式填充到表格中。
优选的，所述模糊更改包括将确定型转化为概率型、将独立型转化为比较型、将整体型转化为局部型以及将数值型转化为范围型。
优选的，在判断填充价值等级为低时，采用防御填充方法进行表格信息填充，所述防御填充方法包括不填充以及采用模糊传递方式进行表格填充。
优选的，所述步骤S4的具体步骤包括：
步骤S41、遍历第一DIKP模型和第二DIKP模型，判断填充的内容是否符合填表人意图；
步骤S42、当填充内容不符合填表人意图时，采用意图覆盖或模糊更改方式对不符合意图的内容进行更改。
与现有技术相比，本发明的有益效果是：
本发明提供了一种面向DIKW内容的意图驱动交互填表方法，对填表人以及表格进行了DIKP模型的构建，其中DIKP模型中包含了填表人以及表格的数据、信息、知识以及意图，通过对比两个DIKP模型的意图并进行价值判断后，可以获得表格的填充价值等级，根据填充价值等级的不同采用降低确定性、模糊传 递以及防御填充的方式进行表格信息填充，最后对填充的信息进行反馈核验，通过将DIKP体系结合到表格自动填写中，充分考虑填表人和表格的意图，实现表格内容填充的智能化以及合理化。
附图说明
为了更清楚地说明本发明实施例中的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的优选实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。
图1为本发明的面向DIKW内容的意图驱动交互填表方法的流程图。
具体实施方式
为了更好理解本发明技术内容，下面提供一具体实施例，并结合附图对本发明做进一步的说明。
参见图1，本发明提供的面向DIKW内容的意图驱动交互填表方法，包括以下步骤：
步骤S1、根据填表人信息构建第一DIKP模型，根据表格信息构建第二DIKP模型。
对于DIKP模型而言，均包括数据体系、信息体系、知识体系以及意图体系，其中数据体系中的数据是由直接贯彻得到的离散元素，在没有上下文的情况下不体现出具体意义，与特定的意图相结合后会产生信息；信息体系中的信息由数据与意图相结合，是对数据按特定意图的回应，是一种有方向性的表达，多个多个意图可以与数据或信息相关，通过将目标数据与至少一个意图联系起来，实现从数据类型到信息类型的概念转换；意图是人类具有的与特定事物相关的隐含或明确的目的或目标，在数据、信息以及知识当中，意图主要与数据进行关联结合，单个数据或多个数据可以与一个或多个意图进行结合。意图本身之间的主要关系为与或非以及包含，一个意图可划分出几个子意图；知识由数据和信息经过结构化形式化的统计和推导演绎获得，在信息的基础上进一步凝练 得到，形成一种知识规则，反应一定的规律性，具有一定的稳定性、可重复利用性，数据、信息、知识三者之间可以相互进行结合以及转换。
优选的，所述步骤S1中的第一DIKP模型的构建过程为：判断系统中是否存在填表人以往的DIKP模型，若不存在，则根据填表人当前输入的数据和意图建立第一DIKP模型；若系统中存在填表人以往的DIKP模型，则利用当前输入的数据和意图对以往的DIKP模型进行补充更新，并获得第一DIKP模型，所述步骤S1中的第二DIKP模型根据表格类别、内部表项、位置关系、拓扑结构以及备选项获取表格整体与各表项的意图，并以此构建第二DIKP模型。
在解决实际问题时，所拥有的数据、意图、信息和知识往往都是不充分、不完整和不精确的，在这种情况下更需要建立体系来进行相互验证和补充，其中对于填表人的第一DIKP模型，根据填表人当前输入的数据和意图来进行构建或更新以往的DIKP模型，针对填表人自身的数据和意图产生建立数据体系和意图体系，结合数据体系和意图体系通过“D+P＝I”建立信息体系，继而构建知识体系，数据、意图、信息以及知识具有包含、继承等关系，随着体系建立或补充能进一步产生级联关系和拓扑结构，最终体系以图谱的方式进行存储，从而获得第一DIKP模型，第二DIKP模型的构建过程与第一DIKP模型的构建过程类似，在构建第二DIKP体系后，可以用于分析与填表人的意图差异，以及是否会对填表人产生不利的影响，以此进行价值判断。
无论是第一DIKP模型还是第二DIKP模型，其DIKP体系内部大体框架均是相同的，在DIKP体系中：1)从初始节点到目标节点经过的边数叫距离，经过的节点数叫深度；2)由部分节点和边组成的称为结构；3)从节点发出的边叫出边，也叫分支，其数量称为该节点的出度，进入节点的边叫入边，其数量称为节点的入度；4)从初始节点到目标节点的距离和深度代表认知距离，深度中所有节点出度和称为偏差系数，若认知距离过长，偏差系数过高，找到目标节点耗费代价会过高，但目标精确度会上升；5)节点出度与其子节点的出度和称为该节点的概率偏差因子，概率偏差因子越大，代表其为目标节点的精确度会下降。6)从某一节点经过出边到达的其它节点称为该节点的子节点，进一步经过子节点出边以及下一级子节点出边能到达的节点称为该节点的联通节点；7)若父节点只存在一条出边到达子节点，且子节点也只存在一条入边，即父节点出度为1， 子节点入度为1，称为级联关系；8)若存在父节点出度为1，子节点出度小于或等于1，且深度超过2，称为偏序关系方式，此类结构称为偏序结构；9)若两个结构进行比较，其初始节点相同，若子节点或距离为3以内的联通节点一定数量以上也是相同的，则差异较小；否则差异较大；10)I体系中节点可由D体系中节点与P体系中节点组合映射，K体系中节点可由D体系、I体系和P体系中节点组合映射。
步骤S2、根据第一DIKP模型以及第二DIKP模型进行意图比较以及价值判断，获取表格的填充价值等级，具体步骤为：将表格意图以及填表人意图进行比较，通过语义分析与DIKP模型比较判断意图间的差异，同时根据填表人意图判断填充表格是否产生收益，根据差异大小以及是否产生差异来判断表格的填充价值等级，其中填充价值等级包括高、一般、低三个等级，在表格的意图与填表人的意图差异较小，且能根据填表人意图判断给填表人带来收益的，即为填充价值等级高；表格的意图与填表人的意图差异较小，且难以根据填表人意图判断是否带来收益的，即为填充价值等级与一般；表格的意图与填表人的意图差异较大的，即为填充价值等级低。
举个例子，当Tom填写企业招募表格，在Tom的第一DIKP模型中Tom的意图是想要入职，且表格意图是招募新员工，通过语义分析和模型比较发现差异小，且能根据Tom意图判断具有收益—填表是Tom入职的途径，则对表格标记为“填充价值等级高”，在填写此表格的身体条件中视力这个表项时，如果遍历Tom的第一DIKP模型得知Tom存在轻度近视，而根据表格的第二DIKP模型分析了解到该表项意图是此工作岗位不招收近视人员，与Tom的意图差异大，则对该表项记为“填充价值等级低”。
步骤S3、在获取表格的填充价值等级后，根据填充价值等级的高低，采用不同的方法进行表格信息填充，其中在填充价值等级为高时，采用降低确定性进行表格信息填充，在填充价值等级为一般时，采用模糊传递方法进行表格信息填充，在填充价值等级为低时，采用防御填充方法进行表格信息填充。
经过意图比较价值判断之后，需要对填充价值等级高的表项进行填充，但是在数据、信息、知识不充分、不完整和不精确的情况下，内容填充的过程中数据和信息流动是具有一种位置不确定性的，数据可以与多个意图结合，得到 的信息是较为发散的，那数据和信息填入的位置也可能是发散的，这称为位置不确定性，这种位置不确定性是难以消除的，因为拥有的数据、信息和知识是不充分、不完整和不精确的，但是通过遍历填表人的第一DIKP模型，找寻所有可能相关的节点，并根据跨体系节点间的映射关系，给可能需要填充的内容节点加上相关的数据、信息或知识将这种不确定性降低。
采用降低确定性方法进行表格信息填充的具体步骤包括：
步骤S311、获取第一DIKP模型中关于填表人的第一数据；
步骤S312、获取第二DIKP模型中的表格与第一数据具有关联的多个表项；
步骤S313、针对步骤S312中具有关联的多个表项遍历第一DIKP模型，并根据映射关系找到与第一数据相关联的信息；
步骤S314、根据与第一数据相关联的信息将第一数据填充到对应的表项中。
例如Tom的数据体系中具有DAT(18)、DAT(5)、DAT(70)这三个数据，在不充分、不完整和不精确的情况下暂不清楚三者具体是什么含义，三者在表格的第二DIKP模型中都可填入年龄、编号等表项，针对DAT(年龄)这个表项，遍历Tom的第一DIKP模型，并根据映射关系找到可能与年龄相关的DIKP内容，假设在信息体系中存在节点INF(Tom,just graduated,from high school)与DAT(18)具有映射关系，那么结合DAT(18)、DAT(5)、DAT(70)三个填表人数据和DAT(Age)这个表项，可以推测出INF(Tom’s age,is,18)，将DAT(18)这个数据填入DAT(Age)当中。
优选的，在判断填充价值等级为一般时，采用模糊传递方法进行表格信息填充，具体步骤为：获取第一DIKP模型中关于填表人的第二数据，对第二数据采用切断数据、信息、知识之间的包含、级联、偏序关系方式或模糊更改的方式填充到表格中。
在面临填充价值等级一般的表格或表项时，填表人可能希望自己的数据信息知识仅仅只能被表格获取到必要的部分，满足意图即可，减少泄露的风险，甚至在一定程度上不想让表格知道某些完整的数据或信息，想要验证表格的真实性和意图，这种情况下数据或信息从填表人到表格的流动可以进行模糊传递，结合填表人的第一DIKP模型，对数据、信息或知识进行某种变化，提高概率偏差因子，或切断数据、信息或知识之间的包含、级联或偏序等关系。
例如，Tom考试考了96分，但是其不想让表格知道具体分数，遍历表格数据体系发现DAT(96)是DAT(high)的联通节点，可以将DAT(96)转化为DAT(high)进行传递，提高概率偏差因子，降低精确度，达到Tom的意图。
模糊传递可以在填充过程中对填充内容进行关键节点的删除，破坏数据、信息或知识传递过去的关系，本发明还提供了模糊更改的方式进行表格信息填充，模糊更改包括将确定型转化为概率型、将独立型转化为比较型、将整体型转化为局部型以及将数值型转化为范围型。
1.确定型到概率型转化：
确定型到概率型的转化是将一个确定的数据或信息以概率的形式表现出来，如INF(Tom被上家单位辞退了)可以转化为INF(Tom可能是主动辞职的，也可能是被辞退了)，这种情况下在填充DAT(工作经历)这个表项不能完全确定Tom的真实去职信息，对填表人的意图PUP(Tom不想让人知道自己是被辞退的)实现了满足，同时又给出可参考信息给表格，一定程度满足表格意图，达到填充要求，概率型数据需要表格根据填表人所填数据、信息或知识并结合自身DIKP体系来进行内容的推理判断，同时也降低了填表人的数据或信息泄露风险。
2.独立型到比较型转化：
独立型到比较型的转化是一种偏序转化，将一个独立表示的数据或信息转变成与其它数据或信息的比较形式表现出来，如INF(Tom身高170cm)，这是一个独立的信息，可以转变为INF(Tom比Jerry高了5cm)这个比较形式的信息。在填充价值等级一般的情况下，难以判断表格得到信息后会对填表人产生什么影响，但如果表格已经从Jerry方得到了INF(Jerry身高是165cm)这个信息，说明表格可能得到了Jerry方的信任，其得到信息后可能同样不会对Tom产生消极影响，那么通过Jerry的信息能知道Tom身高是170cm，否则表格无法获取到INF(Tom身高170cm)这个信息，达到了一种模糊传递的效果。
3.整体型到局部型转化：
整体型到局部型转化是一种去完整性的过程，表明将填表人的某一数据或信息进行部分的切分，不将全部内容提供给对方，而是提供局部的一部分。如Tom的年龄是DAT(29)，但Tom是个女性，不想让人知道她马上要30岁了，可 以转变为DAT(2*)，将另一部分给隐藏，面对某些表格意图如PUP(了解填表人是否成年)，则DAT(2*)这个数据能够满足表格的需求，同时又满足了Tom的意图。
4.数值型到范围型转化：
数值型到范围性转化是将某些数值型数据或信息掩藏在某一类范围当中，根据填表人的意图可以隐藏到不同的范围之中，如Tom体重是80kg，其意图不想让别人知道他有这么重，可以将DAT(80kg)这个数值型数据转变为范围型数据DAT(>60kg)，满足Tom的意图。又或者Tom今年16岁，想要享受一些未成年人便利，如购半价票等，但其不想对陌生人透露准确的年龄信息，可以转变为INF(Tom是个未成年人)，将16岁放入未成年这个范围之内，满足Tom的意图。
优选的，在判断填充价值等级为低时，采用防御填充方法进行表格信息填充，所述防御填充方法包括不填充以及采用模糊传递方式进行表格填充。
此步骤也可以称为误导填充，针对填充价值低的表格或表项，并且难以推测出其隐藏意图的情况下，有理由怀疑其对填表人存在数据信息安全隐患，需要进行一种防御性填充措施，在面对此类表格时，填充价值极低情况下直接不进行填充，避免给填表人带来过多损失，而面对此类表项时，需要计算该表项节点的出入度，若出入度值较高，即使不进行填充，表格也很可能根据其它表项的所填内容与隐藏意图推测出填表人在该表项的填充内容，这种情况下需要进行误导填充，在表格的意图基础上对填充内容进行模糊传递，更改填充内容。
内容填充完毕之前需要对填表人所填入的数据或信息进行内容核验，由于数据、信息、意图以及知识体系中具有一定的包含、级联以及偏序关系，容易在某一个表项节点的填充上违背另一个表项节点的意图，填充的数据或信息关联到了其它表项，因此需要执行步骤S4、对填充完成的表格信息进行反馈核验，具体步骤包括：步骤S41、遍历第一DIKP模型和第二DIKP模型，判断填充的内容是否符合填表人意图；步骤S42、当填充内容不符合填表人意图时，采用意图覆盖或模糊更改方式对不符合意图的内容进行更改。
意图覆盖包括将填表人意图根据包含、级联或偏序关系进行补充覆盖映射，如填表人不想让人知道其年龄，可由数据的级联或偏序关系将意图从DAT(age)补充映射到DAT(birthday)和DAT(ID number)，避免出现在某一个表项的填充上违背填表人其它的意图；2)切断包含、级联或偏序关系，可利用模糊传递中的方 法对待填充数据或信息进行模糊或更改，切断待填充数据或信息间的关系，当填表人不想让人知道其年龄是DAT(29)，填充其出生日期时可进行整体型到局部型转化，隐藏年份，如DAT(5.7)，或进行其它方式的转化，使其不违背填表人的意图。
本发明通过将DIKP体系结合到自动填表中，并根据填表人和表格双方的意图来进行表格信息的填充，最终填充的信息以填表人的意图为驱动，可以保证填写的内容符合填表人的意图，从而达到智能化以及合理化，减少填表人繁琐的手动填表过程，并且对填表人的隐私进行保护。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。

Claims (10)

  1. 面向DIKW内容的意图驱动交互填表方法，其特征在于，包括以下步骤：
    步骤S1、根据填表人信息构建第一DIKP模型，根据表格信息构建第二DIKP模型；
    步骤S2、根据第一DIKP模型以及第二DIKP模型进行意图比较以及价值判断，获取表格的填充价值等级；
    步骤S3、根据填充价值等级的不同采用降低确定性、模糊传递以及防御填充方法进行表格信息填充；
    步骤S4、对填充完成的表格信息进行反馈核验。
  2. 根据权利要求1所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述步骤S1中的第一DIKP模型的构建过程为：判断系统中是否存在填表人以往的DIKP模型，若不存在，则根据填表人当前输入的数据和意图建立第一DIKP模型；若系统中存在填表人以往的DIKP模型，则利用当前输入的数据和意图对以往的DIKP模型进行补充更新，并获得第一DIKP模型。
  3. 根据权利要求1所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述步骤S1中的第二DIKP模型根据表格类别、内部表项、位置关系、拓扑结构以及备选项获取表格整体与各表项的意图，并以此构建第二DIKP模型。
  4. 根据权利要求1所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述步骤S2的具体步骤为：将表格意图以及填表人意图进行比较，通过语义分析与DIKP模型比较判断意图间的差异，同时根据填表人意图判断填充表格是否产生收益，根据差异大小以及是否产生差异来判断表格的填充价值等级。
  5. 根据权利要求4所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述填充价值等级包括高、一般、低三个等级，在表格的意图与填表人的意图差异较小，且能根据填表人意图判断给填表人带来收益的，即为填充价值等级高；表格的意图与填表人的意图差异较小，且难以根据填表人意图判断是否带来收益的，即为填充价值等级与一般；表格的意图与填表人的意图差异较大的，即为填充价值等级低。
  6. 根据权利要求5所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，在判断填充价值等级为高时，采用降低确定性方法进行表格信息填充，具体步骤包括：
    步骤S311、获取第一DIKP模型中关于填表人的第一数据；
    步骤S312、获取第二DIKP模型中的表格与第一数据具有关联的多个表项；
    步骤S313、针对步骤S312中具有关联的多个表项遍历第一DIKP模型，并根据映射关系找到与第一数据相关联的信息；
    步骤S314、根据与第一数据相关联的信息将第一数据填充到对应的表项中。
  7. 根据权利要求5所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，在判断填充价值等级为一般时，采用模糊传递方法进行表格信息填充，具体步骤为：获取第一DIKP模型中关于填表人的第二数据，对第二数据采用切断数据、信息、知识之间的包含、级联、偏序关系方式或模糊更改的方式填充到表格中。
  8. 根据权利要求7所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述模糊更改包括将确定型转化为概率型、将独立型转化为比较型、将整体型转化为局部型以及将数值型转化为范围型。
  9. 根据权利要求5所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，在判断填充价值等级为低时，采用防御填充方法进行表格信息填充，所述防御填充方法包括不填充以及采用模糊传递方式进行表格填充。
  10. 根据权利要求1所述的面向DIKW内容的意图驱动交互填表方法，其特征在于，所述步骤S4的具体步骤包括：
    步骤S41、遍历第一DIKP模型和第二DIKP模型，判断填充的内容是否符合填表人意图；
    步骤S42、当填充内容不符合填表人意图时，采用意图覆盖或模糊更改方式对不符合意图的内容进行更改。
PCT/CN2021/127300 2021-09-01 2021-10-29 面向dikw内容的意图驱动交互填表方法 WO2023029189A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/564,266 US20230065902A1 (en) 2021-09-01 2021-12-29 Intention-driven interactive form-filling method for dikw cotent

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111022025.8 2021-09-01
CN202111022025.8A CN113807063B (zh) 2021-09-01 2021-09-01 面向dikw内容的意图驱动交互填表方法

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/564,266 Continuation US20230065902A1 (en) 2021-09-01 2021-12-29 Intention-driven interactive form-filling method for dikw cotent

Publications (1)

Publication Number Publication Date
WO2023029189A1 true WO2023029189A1 (zh) 2023-03-09

Family

ID=78942163

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/127300 WO2023029189A1 (zh) 2021-09-01 2021-10-29 面向dikw内容的意图驱动交互填表方法

Country Status (2)

Country Link
CN (1) CN113807063B (zh)
WO (1) WO2023029189A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114357192B (zh) * 2021-12-31 2024-04-19 海南大学 基于dikw的内容完整性建模与判断方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104933037A (zh) * 2014-03-20 2015-09-23 无锡伍新网络科技有限公司 个人信息翻译方法及装置
CN110717318A (zh) * 2019-10-10 2020-01-21 海南大学 意图驱动的适应竞争及合作意向的内容填充系统
US20200117708A1 (en) * 2018-10-11 2020-04-16 International Business Machines Corporation Populating spreadsheets using relational information from documents
CN112257396A (zh) * 2020-10-20 2021-01-22 浪潮云信息技术股份公司 一种基于人工智能技术的手机端辅助填表方法
CN113657083A (zh) * 2021-08-30 2021-11-16 海南大学 面向意图计算与推理的dikw资源交互填充系统

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120084098A1 (en) * 2010-09-30 2012-04-05 Andrew Braunstein System and method for populating electronic healthcare forms
CN107688614B (zh) * 2017-08-04 2018-08-10 平安科技（深圳）有限公司 意图获取方法、电子装置及计算机可读存储介质
CN112597285B (zh) * 2020-12-10 2021-08-10 太极计算机股份有限公司 一种基于知识图谱的人机交互方法及系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104933037A (zh) * 2014-03-20 2015-09-23 无锡伍新网络科技有限公司 个人信息翻译方法及装置
US20200117708A1 (en) * 2018-10-11 2020-04-16 International Business Machines Corporation Populating spreadsheets using relational information from documents
CN110717318A (zh) * 2019-10-10 2020-01-21 海南大学 意图驱动的适应竞争及合作意向的内容填充系统
CN112257396A (zh) * 2020-10-20 2021-01-22 浪潮云信息技术股份公司 一种基于人工智能技术的手机端辅助填表方法
CN113657083A (zh) * 2021-08-30 2021-11-16 海南大学 面向意图计算与推理的dikw资源交互填充系统

Also Published As

Publication number Publication date
CN113807063A (zh) 2021-12-17
CN113807063B (zh) 2023-09-05

Similar Documents

Publication Publication Date Title
Hartner et al. Procedural fairness and tax compliance
US8826455B2 (en) Method and apparatus for automated assignment of access permissions to users
Al-Quran et al. Neutrosophic vague soft expert set theory
US9069945B2 (en) User validation in a social network
JP2017535000A (ja) 文書内の潜在的重要事実を自動識別するシステム及び方法
Geng et al. Pythagorean fuzzy uncertain linguistic TODIM method and their application to multiple criteria group decision making
US20030078900A1 (en) Distributed decision processing system with advanced comparison engine
Saylam et al. Conceptualizing citizen-to-citizen (C2C) interactions within the E-government domain
Akram et al. Decision-making methods based on hybrid mF models
CN110727852A (zh) 一种推送招聘推荐服务的方法、装置及终端
CN107122438A (zh) 一种司法案件检索方法及系统
WO2023029189A1 (zh) 面向dikw内容的意图驱动交互填表方法
CN111275395A (zh) 一种基于区块链的去中心化的企业绩效考核方法
CN106997396A (zh) 供需信息互联方法及系统
Wang et al. A three-way decision approach with risk strategies in hesitant fuzzy decision information systems
Ma et al. Multicriteria decision making with cognitive limitations: A DS/AHP‐based approach
US20230065902A1 (en) Intention-driven interactive form-filling method for dikw cotent
Shayeb et al. Assessment of arterial signal timings based on various operational policies and optimization tools
Lv et al. A regret theory-based GEDM method with heterogeneous probabilistic hesitant information and its application to emergency plans selection of COVID-19
Mandal et al. Multiplicative consistency analysis of linguistic preference relation with self‐confidence level and self‐doubting level and its application in a group decision making
WO2023029185A1 (zh) 面向意图计算与推理的dikw资源交互填充系统
US8073717B2 (en) Systems for digital management of underwriting insurance contracts
CN117291722A (zh) 对象管理方法、相关设备及计算机可读介质
Dutta et al. Aggregation of heterogeneously related information with extended geometric Bonferroni mean and its application in group decision making
US20160224941A1 (en) Candidate Fit Engine - a system that finds competitive candidates and ranks the fit level of each candidate to the position's requirements

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21955718

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE