US20130041910A1 - Attribute Entropy as a Signal in Object Normalization - Google Patents
Attribute Entropy as a Signal in Object Normalization Download PDFInfo
- Publication number
- US20130041910A1 US20130041910A1 US13/585,775 US201213585775A US2013041910A1 US 20130041910 A1 US20130041910 A1 US 20130041910A1 US 201213585775 A US201213585775 A US 201213585775A US 2013041910 A1 US2013041910 A1 US 2013041910A1
- Authority
- US
- United States
- Prior art keywords
- fact
- facts
- objects
- attribute
- entropy
- Prior art date
- Legal status (The legal status 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 status listed.)
- Abandoned
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
Definitions
- the disclosed embodiments relate generally to fact databases. More particularly, the disclosed embodiments relate to determining whether two objects are duplicate objects.
- Data is often organized as large collections of objects. When the objects are added over time, there are often problems with data duplication. For example, a collection may include multiple objects that represent the same entity. Objects are duplicate objects if they represent the same entity, even if the information about the entity contained in the objects is different. Duplicate objects increase storage cost, take longer time to process, and confuse the display of information to the user. Duplicate objects can also lead to inaccurate results, such as an inaccurate count of distinct objects.
- Some applications determine whether two objects are duplicate objects by comparing the value of a specific fact, such as the Social Security Number (SSN), the International Standard Book Number (ISBN), or the Universal Product Code (UPC).
- SSN Social Security Number
- ISBN International Standard Book Number
- UPC Universal Product Code
- Some other applications identify whether two objects are duplicate objects by comparing all common facts of the two objects.
- the two objects are determined to be duplicate objects when the number of matching common facts exceeds a threshold.
- This approach is problematic because it does not always give accurate results. For example, the chance that two objects sharing the same gender being duplicates are much lower than that of two objects sharing the same date of birth.
- this approach is both over-inclusive by identifying distinct objects sharing many facts with little indicating value, and under-inclusive by excluding duplicate objects sharing few facts with great indicating value.
- One method for determining whether two objects are duplicate objects is as follows. Common facts of the two objects are identified. Common facts of the two objects are the facts with the same attribute that are associated with the two objects. Values of the common facts are compared to identify matching common facts. Two objects have a matching common fact if they have a fact with the same attribute and the same value. Entropy of the matching common facts are combined (e.g., added) and compared with an entropy threshold. Objects with a sum of entropies exceeding the entropy threshold are determined to be duplicate objects.
- FIG. 1 shows a network, in accordance with a preferred embodiment of the invention.
- FIGS. 2( a )- 2 ( d ) are block diagrams illustrating a data structure for facts within a repository of FIG. 1 in accordance with preferred embodiments of the invention.
- FIG. 2( e ) is a block diagram illustrating an alternate data structure for facts and objects in accordance with preferred embodiments of the invention.
- FIG. 3 is a flowchart of an exemplary method for determining whether two objects are duplicate objects, in accordance with a preferred embodiment of the invention.
- FIGS. 4( a )-( c ) illustrate examples of determining whether two objects are duplicate objects, in accordance with a preferred embodiment of the invention.
- FIG. 1 shows a system architecture 100 adapted to support one embodiment of the invention.
- FIG. 1 shows components used to add facts into, and retrieve facts from a repository 115 .
- the system architecture 100 includes a network 104 , through which any number of document hosts 102 communicate with a data processing system 106 , along with any number of object requesters 152 , 154 .
- Document hosts 102 store documents and provide access to documents.
- a document is comprised of any machine-readable data including any combination of text, graphics, multimedia content, etc.
- a document may be encoded in a markup language, such as Hypertext Markup Language (HTML), i.e., a web page, in an interpreted language (e.g., JavaScript) or in any other computer readable or executable format.
- HTML Hypertext Markup Language
- a document can include one or more hyperlinks to other documents.
- a typical document will include one or more facts within its content.
- a document stored in a document host 102 may be located and/or identified by a Uniform Resource Locator (URL), or Web address, or any other appropriate form of identification and/or location.
- URL Uniform Resource Locator
- a document host 102 is implemented by a computer system, and typically includes a server adapted to communicate over the network 104 via networking protocols (e.g., TCP/IP), as well as application and presentation protocols (e.g., HTTP, HTML, SOAP, D-HTML, Java).
- networking protocols e.g., TCP/IP
- application and presentation protocols e.g., HTTP, HTML, SOAP, D-HTML, Java.
- the documents stored by a host 102 are typically held in a file directory, a database, or other data repository.
- a host 102 can be implemented in any computing device (e.g., from a PDA or personal computer, a workstation, mini-computer, or mainframe, to a cluster or grid of computers), as well as in any processor architecture or operating system.
- FIG. 1 shows components used to manage facts in a fact repository 115 .
- Data processing system 106 includes one or more importers 108 , one or more janitors 110 , a build engine 112 , a service engine 114 , and a fact repository 115 (also called simply a “repository”).
- Importers 108 operate to process documents received from the document hosts, read the data content of documents, and extract facts (as operationally and programmatically defined within the data processing system 106 ) from such documents.
- the importers 108 also determine the subject or subjects with which the facts are associated, and extract such facts into individual items of data, for storage in the fact repository 115 .
- there are different types of importers 108 for different types of documents for example, dependent on the format or document type.
- Janitors 110 operate to process facts extracted by importer 108 .
- This processing can include but is not limited to, data cleansing, object merging, and fact induction.
- Other types of janitors 110 may be implemented, depending on the types of data management functions desired, such as translation, compression, spelling or grammar correction, and the like.
- Various janitors 110 act on facts to normalize attribute names, and values and delete duplicate and near-duplicate facts so an object does not have redundant information. For example, we might find on one page that Britney Spears' birthday is “12/2/1981” while on another page that her date of birth is “Dec. 2, 1981.” Birthday and Date of Birth might both be rewritten as Birthdate by one janitor and then another janitor might notice that 12/2/1981 and Dec. 2, 1981 are different forms of the same date. It would choose the preferred form, remove the other fact and combine the source lists for the two facts. As a result when you look at the source pages for this fact, on some you'll find an exact match of the fact and on others text that is considered to be synonymous with the fact.
- Build engine 112 builds and manages the repository 115 .
- Service engine 114 is an interface for querying the repository 115 .
- Service engine 114 's main function is to process queries, score matching objects, and return them to the caller but it is also used by janitor 110 .
- Repository 115 stores factual information extracted from a plurality of documents that are located on document hosts 102 .
- a document from which a particular fact may be extracted is a source document (or “source”) of that particular fact.
- source a source of a fact includes that fact (or a synonymous fact) within its contents.
- Repository 115 contains one or more facts.
- each fact is associated with exactly one object.
- One implementation for this association includes in each fact an object ID that uniquely identifies the object of the association.
- any number of facts may be associated with an individual object, by including the object ID for that object in the facts.
- objects themselves are not physically stored in the repository 115 , but rather are defined by the set or group of facts with the same associated object ID, as described below. Further details about facts in repository 115 are described below, in relation to FIGS. 2( a )- 2 ( d ).
- the components of the data processing system 106 will be distributed over multiple computers, communicating over a network.
- repository 115 may be deployed over multiple servers.
- the janitors 110 may be located on any number of different computers. For convenience of explanation, however, the components of the data processing system 106 are discussed as though they were implemented on a single computer.
- document hosts 102 are located on data processing system 106 instead of being coupled to data processing system 106 by a network.
- importer 108 may import facts from a database that is a part of or associated with data processing system 106 .
- FIG. 1 also includes components to access repository 115 on behalf of one or more object requesters 152 , 154 .
- Object requesters are entities that request objects from repository 115 .
- Object requesters 152 , 154 may be understood as clients of the system 106 , and can be implemented in any computer device or architecture.
- a first object requester 152 is located remotely from system 106
- a second object requester 154 is located in data processing system 106 .
- the blog may include a reference to an object whose facts are in repository 115 .
- An object requester 152 such as a browser displaying the blog will access data processing system 106 so that the information of the facts associated with the object can be displayed as part of the blog web page.
- janitor 110 or other entity considered to be part of data processing system 106 can function as object requester 154 , requesting the facts of objects from repository 115 .
- FIG. 1 shows that data processing system 106 includes a memory 107 and one or more processors 116 .
- Memory 107 includes importers 108 , janitors 110 , build engine 112 , service engine 114 , and requester 154 , each of which are preferably implemented as instructions stored in memory 107 and executable by processor 116 .
- Memory 107 also includes repository 115 .
- Repository 115 can be stored in a memory of one or more computer systems or in a type of memory such as a disk.
- FIG. 1 also includes a computer readable medium 118 containing, for example, at least one of importers 108 , janitors 110 , build engine 112 , service engine 114 , requester 154 , and at least some portions of repository 115 .
- FIG. 1 also includes a computer readable medium 118 containing, for example, at least one of importers 108 , janitors 110 , build engine 112 , service engine 114 , requester 154 ,
- data processing system 106 also includes one or more input/output devices 120 that allow data to be input and output to and from data processing system 106 .
- data processing system 106 preferably also includes standard software components such as operating systems and the like and further preferably includes standard hardware components not shown in the figure for clarity of example.
- FIG. 2( a ) shows an example format of a data structure for facts within repository 115 , according to some embodiments of the invention.
- the repository 115 includes facts 204 .
- Each fact 204 includes a unique identifier for that fact, such as a fact ID 210 .
- Each fact 204 includes at least an attribute 212 and a value 214 .
- a fact associated with an object representing George Washington may include an attribute of “date of birth” and a value of “Feb. 22, 1732.”
- all facts are stored as alphanumeric characters since they are extracted from web pages.
- facts also can store binary data values.
- Other embodiments, however, may store fact values as mixed types, or in encoded formats.
- each fact is associated with an object ID 209 that identifies the object that the fact describes.
- object ID 209 identifies the object that the fact describes.
- objects are not stored as separate data entities in memory.
- the facts associated with an object contain the same object ID, but no physical object exists.
- objects are stored as data entities in memory, and include references (for example, pointers or IDs) to the facts associated with the object.
- the logical data structure of a fact can take various forms; in general, a fact is represented by a tuple that includes a fact ID, an attribute, a value, and an object ID.
- the storage implementation of a fact can be in any underlying physical data structure.
- FIG. 2( b ) shows an example of facts having respective fact IDs of 10 , 20 , and 30 in repository 115 .
- Facts 10 and 20 are associated with an object identified by object ID “ 1 .”
- Fact 10 has an attribute of “Name” and a value of “China.”
- Fact 20 has an attribute of “Category” and a value of “Country.”
- the object identified by object ID “ 1 ” has a name fact 205 with a value of “China” and a category fact 206 with a value of “Country.”
- Fact 30 208 has an attribute of “Property” and a value of “”Bill Clinton was the 42nd President of the United States from 1993 to 2001.”
- the object identified by object ID “ 2 ” has a property fact with a fact ID of 30 and a value of “Bill Clinton was the 42nd President of the United States from 1993 to 2001.”
- each fact has one attribute and one value.
- the number of facts associated with an object is not limited; thus while only two facts are shown for the “China” object, in practice there may be dozens, even hundreds of facts associated with a given object.
- the value fields of a fact need not be limited in size or content. For example, a fact about the economy of “China” with an attribute of “Economy” would have a value including several paragraphs of text, numbers, perhaps even tables of figures. This content can be formatted, for example, in a markup language. For example, a fact having an attribute “original html” might have a value of the original html text taken from the source web page.
- FIG. 2( b ) shows the explicit coding of object ID, fact ID, attribute, and value
- content of the fact can be implicitly coded as well (e.g., the first field being the object ID, the second field being the fact ID, the third field being the attribute, and the fourth field being the value).
- Other fields include but are not limited to: the language used to state the fact (English, etc.), how important the fact is, the source of the fact, a confidence value for the fact, and so on.
- FIG. 2( c ) shows an example object reference table 210 that is used in some embodiments. Not all embodiments include an object reference table.
- the object reference table 210 functions to efficiently maintain the associations between object IDs and fact IDs. In the absence of an object reference table 210 , it is also possible to find all facts for a given object ID by querying the repository to find all facts with a particular object ID. While FIGS. 2( b ) and 2 ( c ) illustrate the object reference table 210 with explicit coding of object and fact IDs, the table also may contain just the ID values themselves in column or pair-wise arrangements.
- FIG. 2( d ) shows an example of a data structure for facts within repository 115 , according to some embodiments of the invention showing an extended format of facts.
- the fields include an object reference link 216 to another object.
- the object reference link 216 can be an object ID of another object in the repository 115 , or a reference to the location (e.g., table row) for the object in the object reference table 210 .
- the object reference link 216 allows facts to have as values other objects. For example, for an object “United States,” there may be a fact with the attribute of “president” and the value of “George W. Bush,” with “George W. Bush” being an object having its own facts in repository 115 .
- the value field 214 stores the name of the linked object and the link 216 stores the object identifier of the linked object.
- this “president” fact would include the value 214 of “George W. Bush”, and object reference link 216 that contains the object ID for the for “George W. Bush” object.
- facts 204 do not include a link field 216 because the value 214 of a fact 204 may store a link to another object.
- Each fact 204 also may include one or more metrics 218 .
- a metric provides an indication of the some quality of the fact.
- the metrics include a confidence level and an importance level.
- the confidence level indicates the likelihood that the fact is correct.
- the importance level indicates the relevance of the fact to the object, compared to other facts for the same object.
- the importance level may optionally be viewed as a measure of how vital a fact is to an understanding of the entity or concept represented by the object.
- Each fact 204 includes a list of one or more sources 220 that include the fact and from which the fact was extracted.
- Each source may be identified by a Uniform Resource Locator (URL), or Web address, or any other appropriate form of identification and/or location, such as a unique document identifier.
- URL Uniform Resource Locator
- the facts illustrated in FIG. 2( d ) include an agent field 222 that identifies the importer 108 that extracted the fact.
- the importer 108 may be a specialized importer that extracts facts from a specific source (e.g., the pages of a particular web site, or family of web sites) or type of source (e.g., web pages that present factual information in tabular form), or an importer 108 that extracts facts from free text in documents throughout the Web, and so forth.
- a name fact 207 is a fact that conveys a name for the entity or concept represented by the object ID.
- a name fact 207 includes an attribute 224 of “name” and a value, which is the name of the object. For example, for an object representing the country Spain, a name fact would have the value “Spain.”
- a name fact 207 being a special instance of a general fact 204 , includes the same fields as any other fact 204 ; it has an attribute, a value, a fact ID, metrics, sources, etc.
- the attribute 224 of a name fact 207 indicates that the fact is a name fact, and the value is the actual name.
- the name may be a string of characters.
- An object ID may have one or more associated name facts, as many entities or concepts can have more than one name. For example, an object ID representing Spain may have associated name facts conveying the country's common name “Spain” and the official name “Kingdom of Spain.” As another example, an object ID representing the U.S. Patent and Trademark Office may have associated name facts conveying the agency's acronyms “PTO” and “USPTO” as well as the official name “United States Patent and Trademark Office.” If an object does have more than one associated name fact, one of the name facts may be designated as a primary name and other name facts may be designated as secondary names, either implicitly or explicitly.
- a property fact 208 is a fact that conveys a statement about the entity or concept represented by the object ID.
- Property facts are generally used for summary information about an object.
- a property fact 208 being a special instance of a general fact 204 , also includes the same parameters (such as attribute, value, fact ID, etc.) as other facts 204 .
- the attribute field 226 of a property fact 208 indicates that the fact is a property fact (e.g., attribute is “property”) and the value is a string of text that conveys the statement of interest.
- the value of a property fact may be the text string “Bill Clinton was the 42nd President of the United States from 1993 to 2001.”
- Some object IDs may have one or more associated property facts while other objects may have no associated property facts.
- the data structure of the repository 115 may take on other forms. Other fields may be included in facts and some of the fields described above may be omitted.
- each object ID may have additional special facts aside from name facts and property facts, such as facts conveying a type or category (for example, person, place, movie, actor, organization, etc.) for categorizing the entity or concept represented by the object ID.
- an object's name(s) and/or properties may be represented by special records that have a different format than the general facts records 204 .
- null object As described previously, a collection of facts is associated with an object ID of an object.
- An object may become a null or empty object when facts are disassociated from the object.
- a null object can arise in a number of different ways.
- One type of null object is an object that has had all of its facts (including name facts) removed, leaving no facts associated with its object ID.
- Another type of null object is an object that has all of its associated facts other than name facts removed, leaving only its name fact(s).
- the object may be a null object only if all of its associated name facts are removed.
- a null object represents an entity or concept for which the data processing system 106 has no factual information and, as far as the data processing system 106 is concerned, does not exist.
- facts of a null object may be left in the repository 115 , but have their object ID values cleared (or have their importance to a negative value). However, the facts of the null object are treated as if they were removed from the repository 115 . In some other embodiments, facts of null objects are physically removed from repository 115 .
- FIG. 2( e ) is a block diagram illustrating an alternate data structure 290 for facts and objects in accordance with preferred embodiments of the invention.
- an object 290 contains an object ID 292 and references or points to facts 294 .
- Each fact includes a fact ID 295 , an attribute 297 , and a value 299 .
- an object 290 actually exists in memory 107 .
- An entropy is a numerical measure of the amount of information carried by a variable; the more information the variable carries, the more random the variable, and the greater the entropy. Variables with a narrower range of possible values tend to have lower entropies. Variables whose values are not distributed evenly within the possible values tend to have lower entropies.
- An attribute with low entropy will be one which has a low number of possible values or has many possible values but only a few them occur with any frequency. If an attribute has low entropy, the fact that two objects have the same value for that attribute provides little information about whether the two objects are the same entity. An attribute with a higher entropy is more informative than one with a lower entropy because the value of the attribute is more likely to distinguish an object. In other words, a high entropy means the associated attribute is highly useful for distinguishing among objects. As a result, attribute entropy (also called simply entropy) is a good indicator of the importance of matching common facts when determining whether two objects are duplicate objects.
- the chance that two objects of type human are identical if they share the same gender attribute is much less likely than if they share the same date of birth attribute.
- the entropy might be refined to be calculated based on the type of object—in other words, the weight attribute may not be useful for distinguishing among human objects but it may be useful for distinguishing among planet objects.
- the entropy of the event x is the sum, over all possible outcomes i of x, of the product of the probability of outcome i times the log of the probability of i.
- the janitor 110 uses entropy to identify duplicate objects so that the duplicate objects can be merged together.
- FIG. 3 there is shown a flowchart of an exemplary method for determining whether two objects are duplicates according to one embodiment of the present invention.
- the method illustrated in FIG. 3 may be implemented in software, hardware, or a combination of hardware and software.
- the process commences with a set of objects, each having a set of facts including an attribute and a value (also called fact value).
- the set of objects are retrieved by the janitor 110 from the repository 115 .
- the janitor 110 retrieves the set of objects by asking the service engine 114 for the information stored in the repository 115 .
- the janitor 110 needs to reconstruct the objects based on the facts and object information retrieved.
- the janitor 110 determines 310 the entropy for each attribute, and saves the entropy in one of the metrics 218 of each fact with the attribute.
- the janitor 110 asks the service engine 114 for all facts stored in the repository 115 , and saves the attribute and value of each fact in a table.
- the janitor 110 calculates the associated entropy by applying the Shannon formula illustrated above to all values associated with the attribute in the table.
- the janitor 110 then stores the entropy 0.81 in one of the metrics 218 of each of the four facts with attribute gender.
- the entropy can be calculated using logarithm with bases other than two.
- the janitor 110 recomputes the entropy for each attribute every time the repository 115 is modified.
- entropy is calculated based on the number of possible values that the associated attribute could hypothetically have, while in some other embodiments, entropy is calculated based on the number of actual fact values of the associated attribute. In some embodiments, certain attributes are not given an entropy. In some other embodiments, certain attributes are given a predetermined weight in stead of a calculated entropy.
- the janitor 110 selects 320 common facts of the two objects.
- Common facts are facts sharing the same attribute.
- Objects having matching fact values for common fact might be duplicates.
- An example of the two objects is shown in FIG. 4( a ), another example is shown in FIG. 4( b ).
- objects O1 and O2 are duplicate objects representing the same entity, a Mr. Joe M. Henry with nickname J. H. O1 is associated with six facts with the following attributes: name, phone number, type, date of birth, height, and gender. O2 is associated with four facts: name, phone number, date of birth, and height.
- Object O1 in FIG. 4( b ) is the same object as the object O1 shown in FIG. 4( a ).
- Object O3 in FIG. 4( b ) represents a race horse named John Henry.
- O3 is associated with five facts: name, type, date of birth, height, and gender.
- the common facts of objects O1 and O2 as illustrated in FIG. 4( a ) are the facts with the following attributes: name, phone number, date of birth, and height.
- the common facts of objects O1 and O3 in FIG. 4( b ) are the facts with the following attributes: name, type, date of birth, height, and gender.
- the janitor 110 determines 330 a match of the fact value based on a match measure.
- the match measure is designed to distinguish fact values that can be treated as equivalent from fact values that are essentially different.
- the match measure requires the fact values to be identical in order to be a match.
- two fact values can be determined to match by the match measure when they are lexically equivalent, such as “U.S.A.” and “United States.”
- the match measure can be a fuzzy match based on string or tuple similarity measures (e.g., edit distance, Hamming Distance, Levenshtein Distance, Smith-Waterman Distance, Gotoh Distance, Jaro Distance Metric, Dice's Coefficient, Jaccard Coefficient to name a few).
- fuzzy match measure is based on the edit distance between the fact values of the common fact.
- the edit distance is the minimum number of character insertion, deletion, or substitution needed to transform one fact value into the other. This approach is advantageous because typographical errors can be filtered out by the fuzzy match.
- values of a common fact are deemed to match when the edit distance between the two values is no more than a threshold value of 3.
- Fact with attribute name is a common fact of objects O1-O3 as illustrated in FIGS. 4( a )-( b ).
- the value of fact with attribute name associated with objects O1 (“Joe Henry”) and that of O3 (“John Henry”) have an edit distance of two, thus the janitor 110 determines 330 values of common fact name of O1 and O3 to be matched.
- the value of fact with attribute name associated with object O1 (“Joe Henry”) and that of O2 (“J. H.”) have an edit distance of seven, and the janitor 110 determines 330 values of common fact name of O1 and O2 do not match.
- the match measure treats numeric fact values within a range as matched.
- One such match measure deems two numeric fact values to match if the difference between either one of the two fact values and the arithmetic mean of the two fact values is no more than 1% of the arithmetic mean.
- Fact with attribute height is a common fact of objects O1-O3 as illustrated in FIGS. 4( a )-( b ).
- the arithmetic mean of values of fact with attribute height associated with objects O1 and O2 is 5′9.75′′.
- the janitor 110 determines 330 values of common fact heights of O1 and O2 to be matched.
- the arithmetic mean of values of fact with attribute height associated with objects O1 and O3 is 5′7′′.
- the janitor 110 determines 330 values of common fact heights of O1 and O3 to be not matched.
- the match measure can use other measures, such as geometric mean, harmonic mean, and the like.
- the janitor 110 determines 340 whether the two objects are duplicate objects by combining the entropies of matching common facts into one sum of entropies and comparing the sum of entropies with an entropy threshold measure.
- the entropy threshold measure is designed to separate duplicate objects from distinct objects.
- the entropy threshold measure takes into consideration the entropies of common facts that do not match.
- the entropy threshold measure can deem two objects as duplicate objects if the sum of entropies of matching common facts exceeds the sum of entropies of common facts that do not match for a certain threshold value, for example, a threshold value of 5.
- a threshold value of 5 for example, a threshold value of 5.
- the sum of matching entropies of objects O1 and O2 is 15.17
- the sum of entropies of common facts that do not match is 5.26, the entropy of attribute name, as shown in FIG. 4( c ). Because the difference is 9.91, exceeding the threshold value of 5, the janitor 110 correctly determines 340 O1 and O2 to be duplicate objects.
- the sum of matching entropies of objects O1 and O3 is 6.07, and the sum of entropies of common facts that do not match is 11.02, the sum of entropies of attributes type, date of birth, and height, as shown in FIG. 4( c ). Consequently, the janitor 110 correctly determines 340 O1 and O3 to be distinct objects.
- the entropies of common facts of the duplicate objects need to be recomputed.
- the janitor 110 has access to a copy of the table created when the janitor 110 determines 310 the entropy for each attribute.
- the janitor 110 For each common fact of the duplicate objects, the janitor 110 removes the entry corresponding to the common fact associated with one of the two duplicate objects from the table, calculates a new entropy by applying the Shannon formula to all remaining values associated with the attribute of the common fact in the table, and updates the entropy stored in each fact of the attribute to the new entropy.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computational Linguistics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Fuzzy Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A system and method determines whether two objects are duplicate objects. The system and method matches common facts of the two objects based on a match measure, combines the entropies of the matching common facts, and determines whether the two objects are duplicate objects based on the sum of entropies.
Description
- This application is a continuation application of U.S. patent application Ser. No. 11/356,765, filed Feb. 17, 2006, entitled “Attribute Entropy as a Signal in Object Normalization”, which is incorporated by reference herein in its entirety.
- This application is related to the following U.S. Applications all of which are incorporated by reference herein: U.S. patent application Ser. No. 11/357,748 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/342,290 filed on Jan. 27, 2006; U.S. patent application Ser. No. 11/342,293 filed on Jan. 27, 2006; U.S. patent application Ser. No. 11/356,679 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/356,837 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/356,851 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/356,842 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/356,728 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/341,069 filed on Jan. 27, 2006; U.S. patent application Ser. No. 11/356,838 filed on Feb. 17, 2006; U.S. patent application Ser. No. 11/341,907 filed on Jan. 27, 2006; and U.S. patent application Ser. No. 11/342,277 filed on Jan. 27, 2006.
- The disclosed embodiments relate generally to fact databases. More particularly, the disclosed embodiments relate to determining whether two objects are duplicate objects.
- Data is often organized as large collections of objects. When the objects are added over time, there are often problems with data duplication. For example, a collection may include multiple objects that represent the same entity. Objects are duplicate objects if they represent the same entity, even if the information about the entity contained in the objects is different. Duplicate objects increase storage cost, take longer time to process, and confuse the display of information to the user. Duplicate objects can also lead to inaccurate results, such as an inaccurate count of distinct objects.
- Some applications determine whether two objects are duplicate objects by comparing the value of a specific fact, such as the Social Security Number (SSN), the International Standard Book Number (ISBN), or the Universal Product Code (UPC). This approach is effective when all objects contain one of such facts. The specific facts used for comparison are analogous to the primary keys of database tables in relational databases. But for objects built on incomplete information, some objects may not have any of these facts. Also, when values of such facts associated with either of the two objects are inaccurate, this approach treats the two objects as distinct objects even if other facts associated with the two objects indicate that they are duplicate objects. Thus, this approach only determines whether two objects are duplicate objects when both objects include accurate and complete information.
- Some other applications identify whether two objects are duplicate objects by comparing all common facts of the two objects. The two objects are determined to be duplicate objects when the number of matching common facts exceeds a threshold. This approach is problematic because it does not always give accurate results. For example, the chance that two objects sharing the same gender being duplicates are much lower than that of two objects sharing the same date of birth. By treating all facts equally, this approach is both over-inclusive by identifying distinct objects sharing many facts with little indicating value, and under-inclusive by excluding duplicate objects sharing few facts with great indicating value.
- For these reasons, what is needed is a method and system that determines whether two objects built from imperfect information are duplicate objects.
- One method for determining whether two objects are duplicate objects is as follows. Common facts of the two objects are identified. Common facts of the two objects are the facts with the same attribute that are associated with the two objects. Values of the common facts are compared to identify matching common facts. Two objects have a matching common fact if they have a fact with the same attribute and the same value. Entropy of the matching common facts are combined (e.g., added) and compared with an entropy threshold. Objects with a sum of entropies exceeding the entropy threshold are determined to be duplicate objects.
- These features are not the only features of the invention. In view of the drawings, specification, and claims, many additional features and advantages will be apparent.
-
FIG. 1 shows a network, in accordance with a preferred embodiment of the invention. -
FIGS. 2( a)-2(d) are block diagrams illustrating a data structure for facts within a repository ofFIG. 1 in accordance with preferred embodiments of the invention. -
FIG. 2( e) is a block diagram illustrating an alternate data structure for facts and objects in accordance with preferred embodiments of the invention. -
FIG. 3 is a flowchart of an exemplary method for determining whether two objects are duplicate objects, in accordance with a preferred embodiment of the invention. -
FIGS. 4( a)-(c) illustrate examples of determining whether two objects are duplicate objects, in accordance with a preferred embodiment of the invention. - The figures depict various embodiments of the present invention for purposes of illustration only. One skilled in the art will readily recognize from the following discussion that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the invention described herein.
-
FIG. 1 shows asystem architecture 100 adapted to support one embodiment of the invention.FIG. 1 shows components used to add facts into, and retrieve facts from arepository 115. Thesystem architecture 100 includes anetwork 104, through which any number of document hosts 102 communicate with adata processing system 106, along with any number ofobject requesters - Document hosts 102 store documents and provide access to documents. A document is comprised of any machine-readable data including any combination of text, graphics, multimedia content, etc. A document may be encoded in a markup language, such as Hypertext Markup Language (HTML), i.e., a web page, in an interpreted language (e.g., JavaScript) or in any other computer readable or executable format. A document can include one or more hyperlinks to other documents. A typical document will include one or more facts within its content. A document stored in a document host 102 may be located and/or identified by a Uniform Resource Locator (URL), or Web address, or any other appropriate form of identification and/or location. A document host 102 is implemented by a computer system, and typically includes a server adapted to communicate over the
network 104 via networking protocols (e.g., TCP/IP), as well as application and presentation protocols (e.g., HTTP, HTML, SOAP, D-HTML, Java). The documents stored by a host 102 are typically held in a file directory, a database, or other data repository. A host 102 can be implemented in any computing device (e.g., from a PDA or personal computer, a workstation, mini-computer, or mainframe, to a cluster or grid of computers), as well as in any processor architecture or operating system. -
FIG. 1 shows components used to manage facts in afact repository 115.Data processing system 106 includes one ormore importers 108, one ormore janitors 110, abuild engine 112, aservice engine 114, and a fact repository 115 (also called simply a “repository”). Each of the foregoing are implemented, in one embodiment, as software modules (or programs) executed byprocessor 116.Importers 108 operate to process documents received from the document hosts, read the data content of documents, and extract facts (as operationally and programmatically defined within the data processing system 106) from such documents. Theimporters 108 also determine the subject or subjects with which the facts are associated, and extract such facts into individual items of data, for storage in thefact repository 115. In one embodiment, there are different types ofimporters 108 for different types of documents, for example, dependent on the format or document type. -
Janitors 110 operate to process facts extracted byimporter 108. This processing can include but is not limited to, data cleansing, object merging, and fact induction. In one embodiment, there are a number ofdifferent janitors 110 that perform different types of data management operations on the facts. For example, onejanitor 110 may traverse some set of facts in therepository 115 to find duplicate facts (that is, facts that convey the same factual information) and merge them. Anotherjanitor 110 may also normalize facts into standard formats. Anotherjanitor 110 may also remove unwanted facts fromrepository 115, such as facts related to pornographic content. Other types ofjanitors 110 may be implemented, depending on the types of data management functions desired, such as translation, compression, spelling or grammar correction, and the like. -
Various janitors 110 act on facts to normalize attribute names, and values and delete duplicate and near-duplicate facts so an object does not have redundant information. For example, we might find on one page that Britney Spears' birthday is “12/2/1981” while on another page that her date of birth is “Dec. 2, 1981.” Birthday and Date of Birth might both be rewritten as Birthdate by one janitor and then another janitor might notice that 12/2/1981 and Dec. 2, 1981 are different forms of the same date. It would choose the preferred form, remove the other fact and combine the source lists for the two facts. As a result when you look at the source pages for this fact, on some you'll find an exact match of the fact and on others text that is considered to be synonymous with the fact. -
Build engine 112 builds and manages therepository 115.Service engine 114 is an interface for querying therepository 115.Service engine 114's main function is to process queries, score matching objects, and return them to the caller but it is also used byjanitor 110. -
Repository 115 stores factual information extracted from a plurality of documents that are located on document hosts 102. A document from which a particular fact may be extracted is a source document (or “source”) of that particular fact. In other words, a source of a fact includes that fact (or a synonymous fact) within its contents. -
Repository 115 contains one or more facts. In one embodiment, each fact is associated with exactly one object. One implementation for this association includes in each fact an object ID that uniquely identifies the object of the association. In this manner, any number of facts may be associated with an individual object, by including the object ID for that object in the facts. In one embodiment, objects themselves are not physically stored in therepository 115, but rather are defined by the set or group of facts with the same associated object ID, as described below. Further details about facts inrepository 115 are described below, in relation toFIGS. 2( a)-2(d). - It should be appreciated that in practice at least some of the components of the
data processing system 106 will be distributed over multiple computers, communicating over a network. For example,repository 115 may be deployed over multiple servers. As another example, thejanitors 110 may be located on any number of different computers. For convenience of explanation, however, the components of thedata processing system 106 are discussed as though they were implemented on a single computer. - In another embodiment, some or all of document hosts 102 are located on
data processing system 106 instead of being coupled todata processing system 106 by a network. For example,importer 108 may import facts from a database that is a part of or associated withdata processing system 106. -
FIG. 1 also includes components to accessrepository 115 on behalf of one ormore object requesters repository 115.Object requesters system 106, and can be implemented in any computer device or architecture. As shown inFIG. 1 , afirst object requester 152 is located remotely fromsystem 106, while asecond object requester 154 is located indata processing system 106. For example, in a computer system hosting a blog, the blog may include a reference to an object whose facts are inrepository 115. An object requester 152, such as a browser displaying the blog will accessdata processing system 106 so that the information of the facts associated with the object can be displayed as part of the blog web page. As a second example,janitor 110 or other entity considered to be part ofdata processing system 106 can function as object requester 154, requesting the facts of objects fromrepository 115. -
FIG. 1 shows thatdata processing system 106 includes amemory 107 and one ormore processors 116.Memory 107 includesimporters 108,janitors 110, buildengine 112,service engine 114, andrequester 154, each of which are preferably implemented as instructions stored inmemory 107 and executable byprocessor 116.Memory 107 also includesrepository 115.Repository 115 can be stored in a memory of one or more computer systems or in a type of memory such as a disk.FIG. 1 also includes a computerreadable medium 118 containing, for example, at least one ofimporters 108,janitors 110, buildengine 112,service engine 114,requester 154, and at least some portions ofrepository 115.FIG. 1 also includes one or more input/output devices 120 that allow data to be input and output to and fromdata processing system 106. It will be understood thatdata processing system 106 preferably also includes standard software components such as operating systems and the like and further preferably includes standard hardware components not shown in the figure for clarity of example. -
FIG. 2( a) shows an example format of a data structure for facts withinrepository 115, according to some embodiments of the invention. As described above, therepository 115 includesfacts 204. Eachfact 204 includes a unique identifier for that fact, such as afact ID 210. Eachfact 204 includes at least anattribute 212 and avalue 214. For example, a fact associated with an object representing George Washington may include an attribute of “date of birth” and a value of “Feb. 22, 1732.” In one embodiment, all facts are stored as alphanumeric characters since they are extracted from web pages. In another embodiment, facts also can store binary data values. Other embodiments, however, may store fact values as mixed types, or in encoded formats. - As described above, each fact is associated with an
object ID 209 that identifies the object that the fact describes. Thus, each fact that is associated with a same entity (such as George Washington), will have thesame object ID 209. In one embodiment, objects are not stored as separate data entities in memory. In this embodiment, the facts associated with an object contain the same object ID, but no physical object exists. In another embodiment, objects are stored as data entities in memory, and include references (for example, pointers or IDs) to the facts associated with the object. The logical data structure of a fact can take various forms; in general, a fact is represented by a tuple that includes a fact ID, an attribute, a value, and an object ID. The storage implementation of a fact can be in any underlying physical data structure. -
FIG. 2( b) shows an example of facts having respective fact IDs of 10, 20, and 30 inrepository 115.Facts Fact 10 has an attribute of “Name” and a value of “China.”Fact 20 has an attribute of “Category” and a value of “Country.” Thus, the object identified by object ID “1” has aname fact 205 with a value of “China” and acategory fact 206 with a value of “Country.”Fact 30 208 has an attribute of “Property” and a value of “”Bill Clinton was the 42nd President of the United States from 1993 to 2001.” Thus, the object identified by object ID “2” has a property fact with a fact ID of 30 and a value of “Bill Clinton was the 42nd President of the United States from 1993 to 2001.”In the illustrated embodiment, each fact has one attribute and one value. The number of facts associated with an object is not limited; thus while only two facts are shown for the “China” object, in practice there may be dozens, even hundreds of facts associated with a given object. Also, the value fields of a fact need not be limited in size or content. For example, a fact about the economy of “China” with an attribute of “Economy” would have a value including several paragraphs of text, numbers, perhaps even tables of figures. This content can be formatted, for example, in a markup language. For example, a fact having an attribute “original html” might have a value of the original html text taken from the source web page. - Also, while the illustration of
FIG. 2( b) shows the explicit coding of object ID, fact ID, attribute, and value, in practice the content of the fact can be implicitly coded as well (e.g., the first field being the object ID, the second field being the fact ID, the third field being the attribute, and the fourth field being the value). Other fields include but are not limited to: the language used to state the fact (English, etc.), how important the fact is, the source of the fact, a confidence value for the fact, and so on. -
FIG. 2( c) shows an example object reference table 210 that is used in some embodiments. Not all embodiments include an object reference table. The object reference table 210 functions to efficiently maintain the associations between object IDs and fact IDs. In the absence of an object reference table 210, it is also possible to find all facts for a given object ID by querying the repository to find all facts with a particular object ID. WhileFIGS. 2( b) and 2(c) illustrate the object reference table 210 with explicit coding of object and fact IDs, the table also may contain just the ID values themselves in column or pair-wise arrangements. -
FIG. 2( d) shows an example of a data structure for facts withinrepository 115, according to some embodiments of the invention showing an extended format of facts. In this example, the fields include anobject reference link 216 to another object. Theobject reference link 216 can be an object ID of another object in therepository 115, or a reference to the location (e.g., table row) for the object in the object reference table 210. Theobject reference link 216 allows facts to have as values other objects. For example, for an object “United States,” there may be a fact with the attribute of “president” and the value of “George W. Bush,” with “George W. Bush” being an object having its own facts inrepository 115. In some embodiments, thevalue field 214 stores the name of the linked object and thelink 216 stores the object identifier of the linked object. Thus, this “president” fact would include thevalue 214 of “George W. Bush”, and objectreference link 216 that contains the object ID for the for “George W. Bush” object. In some other embodiments,facts 204 do not include alink field 216 because thevalue 214 of afact 204 may store a link to another object. - Each
fact 204 also may include one ormore metrics 218. A metric provides an indication of the some quality of the fact. In some embodiments, the metrics include a confidence level and an importance level. The confidence level indicates the likelihood that the fact is correct. The importance level indicates the relevance of the fact to the object, compared to other facts for the same object. The importance level may optionally be viewed as a measure of how vital a fact is to an understanding of the entity or concept represented by the object. - Each
fact 204 includes a list of one ormore sources 220 that include the fact and from which the fact was extracted. Each source may be identified by a Uniform Resource Locator (URL), or Web address, or any other appropriate form of identification and/or location, such as a unique document identifier. - The facts illustrated in
FIG. 2( d) include anagent field 222 that identifies theimporter 108 that extracted the fact. For example, theimporter 108 may be a specialized importer that extracts facts from a specific source (e.g., the pages of a particular web site, or family of web sites) or type of source (e.g., web pages that present factual information in tabular form), or animporter 108 that extracts facts from free text in documents throughout the Web, and so forth. - Some embodiments include one or more specialized facts, such as a
name fact 207 and aproperty fact 208. Aname fact 207 is a fact that conveys a name for the entity or concept represented by the object ID. Aname fact 207 includes anattribute 224 of “name” and a value, which is the name of the object. For example, for an object representing the country Spain, a name fact would have the value “Spain.” Aname fact 207, being a special instance of ageneral fact 204, includes the same fields as anyother fact 204; it has an attribute, a value, a fact ID, metrics, sources, etc. Theattribute 224 of aname fact 207 indicates that the fact is a name fact, and the value is the actual name. The name may be a string of characters. An object ID may have one or more associated name facts, as many entities or concepts can have more than one name. For example, an object ID representing Spain may have associated name facts conveying the country's common name “Spain” and the official name “Kingdom of Spain.” As another example, an object ID representing the U.S. Patent and Trademark Office may have associated name facts conveying the agency's acronyms “PTO” and “USPTO” as well as the official name “United States Patent and Trademark Office.” If an object does have more than one associated name fact, one of the name facts may be designated as a primary name and other name facts may be designated as secondary names, either implicitly or explicitly. - A
property fact 208 is a fact that conveys a statement about the entity or concept represented by the object ID. Property facts are generally used for summary information about an object. Aproperty fact 208, being a special instance of ageneral fact 204, also includes the same parameters (such as attribute, value, fact ID, etc.) asother facts 204. Theattribute field 226 of aproperty fact 208 indicates that the fact is a property fact (e.g., attribute is “property”) and the value is a string of text that conveys the statement of interest. For example, for the object ID representing Bill Clinton, the value of a property fact may be the text string “Bill Clinton was the 42nd President of the United States from 1993 to 2001.” Some object IDs may have one or more associated property facts while other objects may have no associated property facts. It should be appreciated that the data structures shown inFIGS. 2( a)-2(d) and described above are merely exemplary. The data structure of therepository 115 may take on other forms. Other fields may be included in facts and some of the fields described above may be omitted. Additionally, each object ID may have additional special facts aside from name facts and property facts, such as facts conveying a type or category (for example, person, place, movie, actor, organization, etc.) for categorizing the entity or concept represented by the object ID. In some embodiments, an object's name(s) and/or properties may be represented by special records that have a different format than the general facts records 204. - As described previously, a collection of facts is associated with an object ID of an object. An object may become a null or empty object when facts are disassociated from the object. A null object can arise in a number of different ways. One type of null object is an object that has had all of its facts (including name facts) removed, leaving no facts associated with its object ID. Another type of null object is an object that has all of its associated facts other than name facts removed, leaving only its name fact(s). Alternatively, the object may be a null object only if all of its associated name facts are removed. A null object represents an entity or concept for which the
data processing system 106 has no factual information and, as far as thedata processing system 106 is concerned, does not exist. In some embodiments, facts of a null object may be left in therepository 115, but have their object ID values cleared (or have their importance to a negative value). However, the facts of the null object are treated as if they were removed from therepository 115. In some other embodiments, facts of null objects are physically removed fromrepository 115. -
FIG. 2( e) is a block diagram illustrating analternate data structure 290 for facts and objects in accordance with preferred embodiments of the invention. In this data structure, anobject 290 contains anobject ID 292 and references or points tofacts 294. Each fact includes afact ID 295, anattribute 297, and avalue 299. In this embodiment, anobject 290 actually exists inmemory 107. - An entropy is a numerical measure of the amount of information carried by a variable; the more information the variable carries, the more random the variable, and the greater the entropy. Variables with a narrower range of possible values tend to have lower entropies. Variables whose values are not distributed evenly within the possible values tend to have lower entropies.
- An attribute with low entropy will be one which has a low number of possible values or has many possible values but only a few them occur with any frequency. If an attribute has low entropy, the fact that two objects have the same value for that attribute provides little information about whether the two objects are the same entity. An attribute with a higher entropy is more informative than one with a lower entropy because the value of the attribute is more likely to distinguish an object. In other words, a high entropy means the associated attribute is highly useful for distinguishing among objects. As a result, attribute entropy (also called simply entropy) is a good indicator of the importance of matching common facts when determining whether two objects are duplicate objects.
- For example, the chance that two objects of type human are identical if they share the same gender attribute is much less likely than if they share the same date of birth attribute. Also note that the entropy might be refined to be calculated based on the type of object—in other words, the weight attribute may not be useful for distinguishing among human objects but it may be useful for distinguishing among planet objects.
- Claude E. Shannon defines entropy in terms of a discrete random event x, with
possible states 1 . . . n as: -
- That is, the entropy of the event x is the sum, over all possible outcomes i of x, of the product of the probability of outcome i times the log of the probability of i.
- In one embodiment, the
janitor 110 uses entropy to identify duplicate objects so that the duplicate objects can be merged together. - Referring to
FIG. 3 , there is shown a flowchart of an exemplary method for determining whether two objects are duplicates according to one embodiment of the present invention. The method illustrated inFIG. 3 may be implemented in software, hardware, or a combination of hardware and software. - The flowchart shown in
FIG. 3 will now be described in detail, illustrated by the example inFIGS. 4( a)-(c). The process commences with a set of objects, each having a set of facts including an attribute and a value (also called fact value). The set of objects are retrieved by thejanitor 110 from therepository 115. In one embodiment, thejanitor 110 retrieves the set of objects by asking theservice engine 114 for the information stored in therepository 115. Depending how object information is stored in therepository 115, thejanitor 110 needs to reconstruct the objects based on the facts and object information retrieved. - The
janitor 110 determines 310 the entropy for each attribute, and saves the entropy in one of themetrics 218 of each fact with the attribute. In one embodiment, thejanitor 110 asks theservice engine 114 for all facts stored in therepository 115, and saves the attribute and value of each fact in a table. For each attribute, thejanitor 110 calculates the associated entropy by applying the Shannon formula illustrated above to all values associated with the attribute in the table. For example, assuming there are four facts with attribute gender, three of them have value male, and one has value female, the entropy of the attribute gender will be: H(gender)=p(male)log2(1/p(male))+p(female)log2(1/p(female))=75%*log2(4/3)+25%*log2 4≈0.81. Thejanitor 110 then stores the entropy 0.81 in one of themetrics 218 of each of the four facts with attribute gender. Alternatively, the entropy can be calculated using logarithm with bases other than two. In some embodiments, thejanitor 110 recomputes the entropy for each attribute every time therepository 115 is modified. - Other embodiments may use other measures of entropy. In some embodiments, entropy is calculated based on the number of possible values that the associated attribute could hypothetically have, while in some other embodiments, entropy is calculated based on the number of actual fact values of the associated attribute. In some embodiments, certain attributes are not given an entropy. In some other embodiments, certain attributes are given a predetermined weight in stead of a calculated entropy.
- For two of the set of objects, the
janitor 110 selects 320 common facts of the two objects. Common facts are facts sharing the same attribute. Objects having matching fact values for common fact might be duplicates. An example of the two objects is shown inFIG. 4( a), another example is shown inFIG. 4( b). - As shown in
FIG. 4( a), objects O1 and O2 are duplicate objects representing the same entity, a Mr. Joe M. Henry with nickname J. H. O1 is associated with six facts with the following attributes: name, phone number, type, date of birth, height, and gender. O2 is associated with four facts: name, phone number, date of birth, and height. Object O1 inFIG. 4( b) is the same object as the object O1 shown inFIG. 4( a). Object O3 inFIG. 4( b) represents a race horse named John Henry. O3 is associated with five facts: name, type, date of birth, height, and gender. Among the facts associated with the objects, there are considerable variations in the associated attributes and values. One example of the entropy of each attribute inFIGS. 4( a)-(b) is illustrated inFIG. 4( c). - The common facts of objects O1 and O2 as illustrated in
FIG. 4( a) are the facts with the following attributes: name, phone number, date of birth, and height. The common facts of objects O1 and O3 inFIG. 4( b) are the facts with the following attributes: name, type, date of birth, height, and gender. - For each common fact associated with the two objects, the
janitor 110 determines 330 a match of the fact value based on a match measure. The match measure is designed to distinguish fact values that can be treated as equivalent from fact values that are essentially different. In one example, the match measure requires the fact values to be identical in order to be a match. In another example, two fact values can be determined to match by the match measure when they are lexically equivalent, such as “U.S.A.” and “United States.” Alternatively, the match measure can be a fuzzy match based on string or tuple similarity measures (e.g., edit distance, Hamming Distance, Levenshtein Distance, Smith-Waterman Distance, Gotoh Distance, Jaro Distance Metric, Dice's Coefficient, Jaccard Coefficient to name a few). - One example of a fuzzy match measure is based on the edit distance between the fact values of the common fact. The edit distance is the minimum number of character insertion, deletion, or substitution needed to transform one fact value into the other. This approach is advantageous because typographical errors can be filtered out by the fuzzy match.
- For example, values of a common fact are deemed to match when the edit distance between the two values is no more than a threshold value of 3. Fact with attribute name is a common fact of objects O1-O3 as illustrated in
FIGS. 4( a)-(b). The value of fact with attribute name associated with objects O1 (“Joe Henry”) and that of O3 (“John Henry”) have an edit distance of two, thus thejanitor 110 determines 330 values of common fact name of O1 and O3 to be matched. The value of fact with attribute name associated with object O1 (“Joe Henry”) and that of O2 (“J. H.”) have an edit distance of seven, and thejanitor 110 determines 330 values of common fact name of O1 and O2 do not match. - In another example, the match measure treats numeric fact values within a range as matched. One such match measure deems two numeric fact values to match if the difference between either one of the two fact values and the arithmetic mean of the two fact values is no more than 1% of the arithmetic mean. Fact with attribute height is a common fact of objects O1-O3 as illustrated in
FIGS. 4( a)-(b). The arithmetic mean of values of fact with attribute height associated with objects O1 and O2 is 5′9.75″. Because the difference between the value of fact with attribute height associated with either objects and the mean equals 0.36% of the mean ((5′10″-5′9.75″)/5′9.75″=(5′9.75″-5″9.5″)/5′9.75″=0.25″/5′9.75″≈0.36%), less than the 1% threshold, thejanitor 110 determines 330 values of common fact heights of O1 and O2 to be matched. On the other hand, the arithmetic mean of values of fact with attribute height associated with objects O1 and O3 is 5′7″. Because the difference between the value of fact with attribute height associated with either objects and the mean equals 4.48% of the mean ((5′10″-5′7″)/5′7″=(5′7″-5.4″)/5′7″=3″/5′7″≈4.48%), more than the 1% threshold, thejanitor 110 determines 330 values of common fact heights of O1 and O3 to be not matched. Instead of using arithmetic mean, the match measure can use other measures, such as geometric mean, harmonic mean, and the like. - The
janitor 110 determines 340 whether the two objects are duplicate objects by combining the entropies of matching common facts into one sum of entropies and comparing the sum of entropies with an entropy threshold measure. The entropy threshold measure is designed to separate duplicate objects from distinct objects. - For example, one entropy threshold measure deems two objects as duplicate objects if the sum of entropies is over a threshold value of 7. Assuming the matching common facts of objects O1 and O2 are facts with attributes phone number, date of birth, and height, as illustrated in
FIG. 4( a), the sum of entropies would be 6.32+5.81+3.04=15.17. SeeFIG. 4( c). Because 15.17 exceeds the entropy threshold value 7, thejanitor 110 correctly determines 340 O1 and O2 to be duplicate objects. Assuming the matching common facts of objects O1 and O2 are facts with attributes name and gender, as illustrated inFIG. 4( b), the sum of entropies would be 5.26+0.81=6.07, smaller than the entropy threshold value 7. SeeFIG. 4( c). Consequently, thejanitor 110 correctly determines 340 O1 and O3 to be distinct objects. - In other embodiments, the entropy threshold measure takes into consideration the entropies of common facts that do not match. The entropy threshold measure can deem two objects as duplicate objects if the sum of entropies of matching common facts exceeds the sum of entropies of common facts that do not match for a certain threshold value, for example, a threshold value of 5. Using the example illustrated above, the sum of matching entropies of objects O1 and O2 is 15.17, and the sum of entropies of common facts that do not match is 5.26, the entropy of attribute name, as shown in
FIG. 4( c). Because the difference is 9.91, exceeding the threshold value of 5, thejanitor 110 correctly determines 340 O1 and O2 to be duplicate objects. The sum of matching entropies of objects O1 and O3 is 6.07, and the sum of entropies of common facts that do not match is 11.02, the sum of entropies of attributes type, date of birth, and height, as shown inFIG. 4( c). Consequently, thejanitor 110 correctly determines 340 O1 and O3 to be distinct objects. - To maintain the integrity of entropies, after the
janitor 110 identifies duplicate objects, the entropies of common facts of the duplicate objects need to be recomputed. When re-computing the entropies, only one fact value of the common fact associated with the duplicate objects is used. In one embodiment, thejanitor 110 has access to a copy of the table created when thejanitor 110 determines 310 the entropy for each attribute. For each common fact of the duplicate objects, thejanitor 110 removes the entry corresponding to the common fact associated with one of the two duplicate objects from the table, calculates a new entropy by applying the Shannon formula to all remaining values associated with the attribute of the common fact in the table, and updates the entropy stored in each fact of the attribute to the new entropy. - Finally, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes, and may not have been selected to delineate or circumscribe the inventive subject matter. Accordingly, the disclosure of the present invention is intended to be illustrative, but not limiting, of the scope of the invention, which is set forth in the following claims.
Claims (1)
1. A computer-implemented method of determining if a first object and a second object represent a same entity, the method comprising:
identifying one or more common attributes between the first object and the second object;
determining an entropy for each of the one or more common attributes, wherein a respective entropy is a numeric value measuring an amount of information carried by the respective attribute;
identifying a subset of the one or more common attributes whose respective values are equivalent;
determining whether the first object and the second object represent the same entity by comparing a sum of entropies for the subset of the one or more common attributes to an entropy threshold measure; and
in response to determining that the first object and the second object represent the same entity, merging the first object and the second object.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US13/585,775 US20130041910A1 (en) | 2006-02-17 | 2012-08-14 | Attribute Entropy as a Signal in Object Normalization |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/356,765 US8244689B2 (en) | 2006-02-17 | 2006-02-17 | Attribute entropy as a signal in object normalization |
US13/585,775 US20130041910A1 (en) | 2006-02-17 | 2012-08-14 | Attribute Entropy as a Signal in Object Normalization |
Related Parent Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/356,765 Continuation US8244689B2 (en) | 2004-12-30 | 2006-02-17 | Attribute entropy as a signal in object normalization |
Publications (1)
Publication Number | Publication Date |
---|---|
US20130041910A1 true US20130041910A1 (en) | 2013-02-14 |
Family
ID=38429637
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/356,765 Active 2027-12-10 US8244689B2 (en) | 2004-12-30 | 2006-02-17 | Attribute entropy as a signal in object normalization |
US13/585,775 Abandoned US20130041910A1 (en) | 2006-02-17 | 2012-08-14 | Attribute Entropy as a Signal in Object Normalization |
Family Applications Before (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US11/356,765 Active 2027-12-10 US8244689B2 (en) | 2004-12-30 | 2006-02-17 | Attribute entropy as a signal in object normalization |
Country Status (1)
Country | Link |
---|---|
US (2) | US8244689B2 (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104899189A (en) * | 2015-05-27 | 2015-09-09 | 深圳市华傲数据技术有限公司 | Object name matching method based on information entropy |
WO2015183306A1 (en) * | 2014-05-30 | 2015-12-03 | Hewlett-Packard Development Company, L.P. | Processing a table of columns |
Families Citing this family (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7529728B2 (en) | 2003-09-23 | 2009-05-05 | Salesforce.Com, Inc. | Query optimization in a multi-tenant database system |
US8543566B2 (en) | 2003-09-23 | 2013-09-24 | Salesforce.Com, Inc. | System and methods of improving a multi-tenant database query using contextual knowledge about non-homogeneously distributed tenant data |
US8095531B2 (en) | 2006-10-03 | 2012-01-10 | Salesforce.Com, Inc. | Methods and systems for controlling access to custom objects in a database |
US20080103902A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Orchestration and/or exploration of different advertising channels in a federated advertising network |
US20080103897A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Normalizing and tracking user attributes for transactions in an advertising exchange |
US20080103955A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Accounting for trusted participants in an online advertising exchange |
US20080103795A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Lightweight and heavyweight interfaces to federated advertising marketplace |
US8589233B2 (en) * | 2006-10-25 | 2013-11-19 | Microsoft Corporation | Arbitrage broker for online advertising exchange |
US20080103953A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Tool for optimizing advertising across disparate advertising networks |
US8533049B2 (en) * | 2006-10-25 | 2013-09-10 | Microsoft Corporation | Value add broker for federated advertising exchange |
US8788343B2 (en) * | 2006-10-25 | 2014-07-22 | Microsoft Corporation | Price determination and inventory allocation based on spot and futures markets in future site channels for online advertising |
US20080103952A1 (en) * | 2006-10-25 | 2008-05-01 | Microsoft Corporation | Specifying and normalizing utility functions of participants in an advertising exchange |
US8862980B2 (en) * | 2007-06-29 | 2014-10-14 | Business Objects Software Limited | Guided graphical navigation through multidimensional databases |
US9229971B2 (en) * | 2010-12-21 | 2016-01-05 | Business Objects Software Limited | Matching data based on numeric difference |
US8560468B1 (en) * | 2011-02-10 | 2013-10-15 | Google Inc. | Learning expected values for facts |
EP2492824B1 (en) * | 2011-02-23 | 2020-04-01 | Harman Becker Automotive Systems GmbH | Method of searching a data base, navigation device and method of generating an index structure |
US10540354B2 (en) * | 2011-10-17 | 2020-01-21 | Micro Focus Llc | Discovering representative composite CI patterns in an it system |
US9305150B2 (en) * | 2012-12-10 | 2016-04-05 | Lookout, Inc. | Method and system for managing user login behavior on an electronic device for enhanced security |
US9552360B2 (en) | 2013-09-27 | 2017-01-24 | International Business Machines Corporation | Resource reconciliation based on external factors |
US10394922B2 (en) * | 2014-06-30 | 2019-08-27 | Facebook, Inc. | User index with nicknames |
US20160125005A1 (en) * | 2014-10-30 | 2016-05-05 | The Johns Hopkins University | Apparatus and Method for Profiling Activities and Transitions |
CA3072192A1 (en) | 2017-08-11 | 2019-02-14 | Ancestry.Com Operations Inc. | Diversity evaluation in genealogy search |
US10417335B2 (en) * | 2017-10-10 | 2019-09-17 | Colossio, Inc. | Automated quantitative assessment of text complexity |
US10831927B2 (en) * | 2017-11-22 | 2020-11-10 | International Business Machines Corporation | Noise propagation-based data anonymization |
CN107729571B (en) * | 2017-11-23 | 2020-04-14 | 北京天广汇通科技有限公司 | Relationship discovery method and device |
EP3918486A4 (en) | 2019-02-01 | 2022-10-19 | Ancestry.com Operations Inc. | Search and ranking of records across different databases |
US11354297B2 (en) * | 2019-08-07 | 2022-06-07 | International Business Machines Corporation | Detecting positivity violations in multidimensional data |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020169730A1 (en) * | 2001-08-29 | 2002-11-14 | Emmanuel Lazaridis | Methods for classifying objects and identifying latent classes |
US20030177110A1 (en) * | 2002-03-15 | 2003-09-18 | Fujitsu Limited | Profile information recommendation method, program and apparatus |
US20060047838A1 (en) * | 2004-06-25 | 2006-03-02 | Abhishek Chauhan | Inferring server state in a stateless communication protocol |
Family Cites Families (172)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5010478A (en) * | 1986-04-11 | 1991-04-23 | Deran Roger L | Entity-attribute value database system with inverse attribute for selectively relating two different entities |
US5133075A (en) * | 1988-12-19 | 1992-07-21 | Hewlett-Packard Company | Method of monitoring changes in attribute values of object in an object-oriented database |
US5440730A (en) * | 1990-08-09 | 1995-08-08 | Bell Communications Research, Inc. | Time index access structure for temporal databases having concurrent multiple versions |
CA2048306A1 (en) | 1990-10-02 | 1992-04-03 | Steven P. Miller | Distributed configuration profile for computing system |
US5347653A (en) | 1991-06-28 | 1994-09-13 | Digital Equipment Corporation | System for reconstructing prior versions of indexes using records indicating changes between successive versions of the indexes |
US5694590A (en) | 1991-09-27 | 1997-12-02 | The Mitre Corporation | Apparatus and method for the detection of security violations in multilevel secure databases |
US5574898A (en) | 1993-01-08 | 1996-11-12 | Atria Software, Inc. | Dynamic software version auditor which monitors a process to provide a list of objects that are accessed |
US7082426B2 (en) * | 1993-06-18 | 2006-07-25 | Cnet Networks, Inc. | Content aggregation method and apparatus for an on-line product catalog |
US5519608A (en) * | 1993-06-24 | 1996-05-21 | Xerox Corporation | Method for extracting from a text corpus answers to questions stated in natural language by using linguistic analysis and hypothesis generation |
US5560005A (en) | 1994-02-25 | 1996-09-24 | Actamed Corp. | Methods and systems for object-based relational distributed databases |
US5680622A (en) | 1994-06-30 | 1997-10-21 | Borland International, Inc. | System and methods for quickly detecting shareability of symbol and type information in header files |
US5608903A (en) | 1994-12-15 | 1997-03-04 | Novell, Inc. | Method and apparatus for moving subtrees in a distributed network directory |
US5717911A (en) * | 1995-01-23 | 1998-02-10 | Tandem Computers, Inc. | Relational database system and method with high availability compliation of SQL programs |
US5793966A (en) * | 1995-12-01 | 1998-08-11 | Vermeer Technologies, Inc. | Computer system and computer-implemented process for creation and maintenance of online services |
US5717951A (en) * | 1995-08-07 | 1998-02-10 | Yabumoto; Kan W. | Method for storing and retrieving information on a magnetic storage medium via data blocks of variable sizes |
US5838979A (en) | 1995-10-31 | 1998-11-17 | Peritus Software Services, Inc. | Process and tool for scalable automated data field replacement |
US5701470A (en) | 1995-12-08 | 1997-12-23 | Sun Microsystems, Inc. | System and method for space efficient object locking using a data subarray and pointers |
US5815415A (en) | 1996-01-19 | 1998-09-29 | Bentley Systems, Incorporated | Computer system for portable persistent modeling |
US5802299A (en) | 1996-02-13 | 1998-09-01 | Microtouch Systems, Inc. | Interactive system for authoring hypertext document collections |
US5778378A (en) * | 1996-04-30 | 1998-07-07 | International Business Machines Corporation | Object oriented information retrieval framework mechanism |
US5920859A (en) * | 1997-02-05 | 1999-07-06 | Idd Enterprises, L.P. | Hypertext document retrieval system and method |
US5819210A (en) | 1996-06-21 | 1998-10-06 | Xerox Corporation | Method of lazy contexted copying during unification |
US6052693A (en) * | 1996-07-02 | 2000-04-18 | Harlequin Group Plc | System for assembling large databases through information extracted from text sources |
US5987460A (en) | 1996-07-05 | 1999-11-16 | Hitachi, Ltd. | Document retrieval-assisting method and system for the same and document retrieval service using the same with document frequency and term frequency |
US5819265A (en) | 1996-07-12 | 1998-10-06 | International Business Machines Corporation | Processing names in a text |
US5787413A (en) * | 1996-07-29 | 1998-07-28 | International Business Machines Corporation | C++ classes for a digital library |
US6820093B2 (en) | 1996-07-30 | 2004-11-16 | Hyperphrase Technologies, Llc | Method for verifying record code prior to an action based on the code |
US5826258A (en) | 1996-10-02 | 1998-10-20 | Junglee Corporation | Method and apparatus for structuring the querying and interpretation of semistructured information |
US6285999B1 (en) | 1997-01-10 | 2001-09-04 | The Board Of Trustees Of The Leland Stanford Junior University | Method for node ranking in a linked database |
US6134555A (en) | 1997-03-10 | 2000-10-17 | International Business Machines Corporation | Dimension reduction using association rules for data mining application |
US5822743A (en) | 1997-04-08 | 1998-10-13 | 1215627 Ontario Inc. | Knowledge-based information retrieval system |
US5974254A (en) | 1997-06-06 | 1999-10-26 | National Instruments Corporation | Method for detecting differences between graphical programs |
US5909689A (en) * | 1997-09-18 | 1999-06-01 | Sony Corporation | Automatic update of file versions for files shared by several computers which record in respective file directories temporal information for indicating when the files have been created |
US6018741A (en) * | 1997-10-22 | 2000-01-25 | International Business Machines Corporation | Method and system for managing objects in a dynamic inheritance tree |
US6112210A (en) * | 1997-10-31 | 2000-08-29 | Oracle Corporation | Apparatus and method for null representation in database object storage |
US5943670A (en) * | 1997-11-21 | 1999-08-24 | International Business Machines Corporation | System and method for categorizing objects in combined categories |
US6349275B1 (en) * | 1997-11-24 | 2002-02-19 | International Business Machines Corporation | Multiple concurrent language support system for electronic catalogue using a concept based knowledge representation |
US6212526B1 (en) * | 1997-12-02 | 2001-04-03 | Microsoft Corporation | Method for apparatus for efficient mining of classification models from databases |
US6094650A (en) | 1997-12-15 | 2000-07-25 | Manning & Napier Information Services | Database analysis using a probabilistic ontology |
US6044366A (en) | 1998-03-16 | 2000-03-28 | Microsoft Corporation | Use of the UNPIVOT relational operator in the efficient gathering of sufficient statistics for data mining |
US6078918A (en) * | 1998-04-02 | 2000-06-20 | Trivada Corporation | Online predictive memory |
US6112203A (en) | 1998-04-09 | 2000-08-29 | Altavista Company | Method for ranking documents in a hyperlinked environment using connectivity and selective content analysis |
US6122647A (en) | 1998-05-19 | 2000-09-19 | Perspecta, Inc. | Dynamic generation of contextual links in hypertext documents |
US6742003B2 (en) * | 2001-04-30 | 2004-05-25 | Microsoft Corporation | Apparatus and accompanying methods for visualizing clusters of data and hierarchical cluster classifications |
US6327574B1 (en) | 1998-07-07 | 2001-12-04 | Encirq Corporation | Hierarchical models of consumer attributes for targeting content in a privacy-preserving manner |
US6240546B1 (en) * | 1998-07-24 | 2001-05-29 | International Business Machines Corporation | Identifying date fields for runtime year 2000 system solution process, method and article of manufacture |
US6665837B1 (en) | 1998-08-10 | 2003-12-16 | Overture Services, Inc. | Method for identifying related pages in a hyperlinked database |
US6694482B1 (en) * | 1998-09-11 | 2004-02-17 | Sbc Technology Resources, Inc. | System and methods for an architectural framework for design of an adaptive, personalized, interactive content delivery system |
US6572661B1 (en) * | 1999-01-11 | 2003-06-03 | Cisco Technology, Inc. | System and method for automated annotation of files |
US6377943B1 (en) * | 1999-01-20 | 2002-04-23 | Oracle Corp. | Initial ordering of tables for database queries |
US7003719B1 (en) * | 1999-01-25 | 2006-02-21 | West Publishing Company, Dba West Group | System, method, and software for inserting hyperlinks into documents |
US6574635B2 (en) * | 1999-03-03 | 2003-06-03 | Siebel Systems, Inc. | Application instantiation based upon attributes and values stored in a meta data repository, including tiering of application layers objects and components |
US6584464B1 (en) | 1999-03-19 | 2003-06-24 | Ask Jeeves, Inc. | Grammar template query system |
US6397228B1 (en) * | 1999-03-31 | 2002-05-28 | Verizon Laboratories Inc. | Data enhancement techniques |
US6263328B1 (en) * | 1999-04-09 | 2001-07-17 | International Business Machines Corporation | Object oriented query model and process for complex heterogeneous database queries |
US6721713B1 (en) * | 1999-05-27 | 2004-04-13 | Andersen Consulting Llp | Business alliance identification in a web architecture framework |
US6606625B1 (en) * | 1999-06-03 | 2003-08-12 | University Of Southern California | Wrapper induction by hierarchical data analysis |
US6711585B1 (en) * | 1999-06-15 | 2004-03-23 | Kanisa Inc. | System and method for implementing a knowledge management system |
US6473898B1 (en) | 1999-07-06 | 2002-10-29 | Pcorder.Com, Inc. | Method for compiling and selecting data attributes |
CA2281331A1 (en) * | 1999-09-03 | 2001-03-03 | Cognos Incorporated | Database management system |
US6754873B1 (en) * | 1999-09-20 | 2004-06-22 | Google Inc. | Techniques for finding related hyperlinked documents using link-based analysis |
GB2371901B (en) * | 1999-09-21 | 2004-06-23 | Andrew E Borthwick | A probabilistic record linkage model derived from training data |
WO2001027713A2 (en) | 1999-10-15 | 2001-04-19 | Milind Kotwal | Method of categorization and indexing of information |
US6665666B1 (en) | 1999-10-26 | 2003-12-16 | International Business Machines Corporation | System, method and program product for answering questions using a search engine |
US6850896B1 (en) * | 1999-10-28 | 2005-02-01 | Market-Touch Corporation | Method and system for managing and providing sales data using world wide web |
US6804667B1 (en) | 1999-11-30 | 2004-10-12 | Ncr Corporation | Filter for checking for duplicate entries in database |
US6963867B2 (en) | 1999-12-08 | 2005-11-08 | A9.Com, Inc. | Search query processing to provide category-ranked presentation of search results |
US7305380B1 (en) | 1999-12-15 | 2007-12-04 | Google Inc. | Systems and methods for performing in-context searching |
US6865582B2 (en) * | 2000-01-03 | 2005-03-08 | Bechtel Bwxt Idaho, Llc | Systems and methods for knowledge discovery in spatial data |
US6606659B1 (en) * | 2000-01-28 | 2003-08-12 | Websense, Inc. | System and method for controlling access to internet sites |
US6665659B1 (en) | 2000-02-01 | 2003-12-16 | James D. Logan | Methods and apparatus for distributing and using metadata via the internet |
US6567936B1 (en) * | 2000-02-08 | 2003-05-20 | Microsoft Corporation | Data clustering using error-tolerant frequent item sets |
AU2001241564A1 (en) * | 2000-02-17 | 2001-08-27 | E-Numerate Solutions, Inc. | Rdl search engine |
US6901403B1 (en) * | 2000-03-02 | 2005-05-31 | Quovadx, Inc. | XML presentation of general-purpose data sources |
US6311194B1 (en) | 2000-03-15 | 2001-10-30 | Taalee, Inc. | System and method for creating a semantic web and its applications in browsing, searching, profiling, personalization and advertising |
US6502102B1 (en) | 2000-03-27 | 2002-12-31 | Accenture Llp | System, method and article of manufacture for a table-driven automated scripting architecture |
US6957213B1 (en) | 2000-05-17 | 2005-10-18 | Inquira, Inc. | Method of utilizing implicit references to answer a query |
EP1156430A2 (en) * | 2000-05-17 | 2001-11-21 | Matsushita Electric Industrial Co., Ltd. | Information retrieval system |
US7062483B2 (en) * | 2000-05-18 | 2006-06-13 | Endeca Technologies, Inc. | Hierarchical data-driven search and navigation system and method for information retrieval |
US6745189B2 (en) * | 2000-06-05 | 2004-06-01 | International Business Machines Corporation | System and method for enabling multi-indexing of objects |
US6963876B2 (en) | 2000-06-05 | 2005-11-08 | International Business Machines Corporation | System and method for searching extended regular expressions |
US20020042707A1 (en) * | 2000-06-19 | 2002-04-11 | Gang Zhao | Grammar-packaged parsing |
US7162499B2 (en) * | 2000-06-21 | 2007-01-09 | Microsoft Corporation | Linked value replication |
MXPA03000110A (en) * | 2000-06-22 | 2006-06-08 | Mayer Yaron | System and method for searching, finding and contacting dates on the internet in instant messaging networks. |
US7080085B1 (en) * | 2000-07-12 | 2006-07-18 | International Business Machines Corporation | System and method for ensuring referential integrity for heterogeneously scoped references in an information management system |
US6728728B2 (en) * | 2000-07-24 | 2004-04-27 | Israel Spiegler | Unified binary model and methodology for knowledge representation and for data and information mining |
US7146536B2 (en) | 2000-08-04 | 2006-12-05 | Sun Microsystems, Inc. | Fact collection for product knowledge management |
US7080073B1 (en) * | 2000-08-18 | 2006-07-18 | Firstrain, Inc. | Method and apparatus for focused crawling |
IT1318683B1 (en) | 2000-08-22 | 2003-08-27 | Ausimont Spa | MIXTURES OF FLUORINATED AND ACRYLIC ELASTOMERS. |
US6556991B1 (en) * | 2000-09-01 | 2003-04-29 | E-Centives, Inc. | Item name normalization |
US6823495B1 (en) | 2000-09-14 | 2004-11-23 | Microsoft Corporation | Mapping tool graphical user interface |
US7143099B2 (en) | 2001-02-08 | 2006-11-28 | Amdocs Software Systems Limited | Historical data warehousing system |
US6820081B1 (en) | 2001-03-19 | 2004-11-16 | Attenex Corporation | System and method for evaluating a structured message store for message redundancy |
US20020147738A1 (en) | 2001-04-06 | 2002-10-10 | Reader Scot A. | Method and appratus for finding patent-relevant web documents |
KR100860354B1 (en) | 2001-04-12 | 2008-09-25 | 코닌클리케 필립스 일렉트로닉스 엔.브이. | Method and system for registering a user preference, and computer readible medium containing computer program product |
US20020169770A1 (en) | 2001-04-27 | 2002-11-14 | Kim Brian Seong-Gon | Apparatus and method that categorize a collection of documents into a hierarchy of categories that are defined by the collection of documents |
US7020662B2 (en) * | 2001-05-29 | 2006-03-28 | Sun Microsystems, Inc. | Method and system for determining a directory entry's class of service based on the value of a specifier in the entry |
US7925616B2 (en) | 2001-06-19 | 2011-04-12 | Microstrategy, Incorporated | Report system and method using context-sensitive prompt objects |
BR0210589A (en) * | 2001-06-22 | 2005-04-26 | Nosa Omoigui | System and method for knowledge retrieval, management, delivery and presentation |
EP1429258A4 (en) * | 2001-07-26 | 2007-08-29 | Ibm | Data processing method, data processing system, and program |
AU2002326118A1 (en) | 2001-08-14 | 2003-03-03 | Quigo Technologies, Inc. | System and method for extracting content for submission to a search engine |
US7386832B2 (en) * | 2001-08-31 | 2008-06-10 | Siebel Systems, Inc. | Configurator using structure to provide a user interface |
US7058653B2 (en) * | 2001-09-17 | 2006-06-06 | Ricoh Company, Ltd. | Tree system diagram output method, computer program and recording medium |
US7020641B2 (en) * | 2001-10-22 | 2006-03-28 | Sun Microsystems, Inc. | Method, system, and program for maintaining a database of data objects |
US7197449B2 (en) * | 2001-10-30 | 2007-03-27 | Intel Corporation | Method for extracting name entities and jargon terms using a suffix tree data structure |
JP3931214B2 (en) * | 2001-12-17 | 2007-06-13 | 日本アイ・ビー・エム株式会社 | Data analysis apparatus and program |
US7421660B2 (en) * | 2003-02-04 | 2008-09-02 | Cataphora, Inc. | Method and apparatus to visually present discussions for data mining purposes |
EP1485825A4 (en) | 2002-02-04 | 2008-03-19 | Cataphora Inc | A method and apparatus for sociological data mining |
US20030149567A1 (en) * | 2002-02-04 | 2003-08-07 | Tony Schmitz | Method and system for using natural language in computer resource utilization analysis via a communications network |
US7436887B2 (en) | 2002-02-06 | 2008-10-14 | Playtex Products, Inc. | Method and apparatus for video frame sequence-based object tracking |
US20030154071A1 (en) * | 2002-02-11 | 2003-08-14 | Shreve Gregory M. | Process for the document management and computer-assisted translation of documents utilizing document corpora constructed by intelligent agents |
US6857053B2 (en) | 2002-04-10 | 2005-02-15 | International Business Machines Corporation | Method, system, and program for backing up objects by creating groups of objects |
US6963880B1 (en) | 2002-05-10 | 2005-11-08 | Oracle International Corporation | Schema evolution of complex objects |
US7003522B1 (en) * | 2002-06-24 | 2006-02-21 | Microsoft Corporation | System and method for incorporating smart tags in online content |
US20040003067A1 (en) * | 2002-06-27 | 2004-01-01 | Daniel Ferrin | System and method for enabling a user interface with GUI meta data |
GB0215464D0 (en) * | 2002-07-04 | 2002-08-14 | Hewlett Packard Co | Combining data descriptions |
CN1316419C (en) * | 2002-08-22 | 2007-05-16 | 新加坡科技研究局 | Prediction by collective likelihood from emerging patterns |
US20040064447A1 (en) * | 2002-09-27 | 2004-04-01 | Simske Steven J. | System and method for management of synonymic searching |
US7096217B2 (en) * | 2002-10-31 | 2006-08-22 | International Business Machines Corporation | Global query correlation attributes |
US20040167910A1 (en) | 2002-12-06 | 2004-08-26 | Attensity Corporation | Integrated data products of processes of integrating mixed format data |
US7277879B2 (en) | 2002-12-17 | 2007-10-02 | Electronic Data Systems Corporation | Concept navigation in data storage systems |
US7181450B2 (en) * | 2002-12-18 | 2007-02-20 | International Business Machines Corporation | Method, system, and program for use of metadata to create multidimensional cubes in a relational database |
US7107528B2 (en) * | 2002-12-20 | 2006-09-12 | International Business Machines Corporation | Automatic completion of dates |
US7472182B1 (en) | 2002-12-31 | 2008-12-30 | Emc Corporation | Data collection policy for storage devices |
GB0304639D0 (en) * | 2003-02-28 | 2003-04-02 | Kiq Ltd | Classification using re-sampling of probability estimates |
US7051023B2 (en) * | 2003-04-04 | 2006-05-23 | Yahoo! Inc. | Systems and methods for generating concept units from search queries |
WO2004092881A2 (en) | 2003-04-07 | 2004-10-28 | Sevenecho, Llc | Method, system and software for digital media narrative personalization |
US7747571B2 (en) | 2003-04-15 | 2010-06-29 | At&T Intellectual Property, I,L.P. | Methods, systems, and computer program products for implementing logical and physical data models |
JP2004362223A (en) | 2003-06-04 | 2004-12-24 | Hitachi Ltd | Information mining system |
US7836391B2 (en) | 2003-06-10 | 2010-11-16 | Google Inc. | Document search engine including highlighting of confident results |
US9026901B2 (en) | 2003-06-20 | 2015-05-05 | International Business Machines Corporation | Viewing annotations across multiple applications |
US7162473B2 (en) | 2003-06-26 | 2007-01-09 | Microsoft Corporation | Method and system for usage analyzer that determines user accessed sources, indexes data subsets, and associated metadata, processing implicit queries based on potential interest to users |
CN1856790A (en) | 2003-07-22 | 2006-11-01 | 基诺技术公司 | Information access using ontologies |
US7496560B2 (en) * | 2003-09-23 | 2009-02-24 | Amazon Technologies, Inc. | Personalized searchable library with highlighting capabilities |
US20070073768A1 (en) * | 2003-10-15 | 2007-03-29 | Goradia Gautam D | Interactive system for building and sharing one's own databank of wisdom bytes, such as words of wisdom, basic truths and/or facts and and feats, in one or more languages |
KR100533810B1 (en) * | 2003-10-16 | 2005-12-07 | 한국전자통신연구원 | Semi-Automatic Construction Method for Knowledge of Encyclopedia Question Answering System |
GB0325626D0 (en) * | 2003-11-03 | 2003-12-10 | Infoshare Ltd | Data aggregation |
US7512553B2 (en) * | 2003-12-05 | 2009-03-31 | International Business Machines Corporation | System for automated part-number mapping |
US20050149851A1 (en) * | 2003-12-31 | 2005-07-07 | Google Inc. | Generating hyperlinks and anchor text in HTML and non-HTML documents |
US8150824B2 (en) * | 2003-12-31 | 2012-04-03 | Google Inc. | Systems and methods for direct navigation to specific portion of target document |
US7424467B2 (en) | 2004-01-26 | 2008-09-09 | International Business Machines Corporation | Architecture for an indexer with fixed width sort and variable width sort |
US7836083B2 (en) | 2004-02-20 | 2010-11-16 | Factiva, Inc. | Intelligent search and retrieval system and method |
US7725498B2 (en) | 2004-04-22 | 2010-05-25 | International Business Machines Corporation | Techniques for identifying mergeable data |
US20060036504A1 (en) * | 2004-08-11 | 2006-02-16 | Allocca William W | Dynamically classifying items for international delivery |
US7809695B2 (en) * | 2004-08-23 | 2010-10-05 | Thomson Reuters Global Resources | Information retrieval systems with duplicate document detection and presentation functions |
US20060053175A1 (en) * | 2004-09-03 | 2006-03-09 | Biowisdom Limited | System and method for creating, editing, and utilizing one or more rules for multi-relational ontology creation and maintenance |
US20060053171A1 (en) * | 2004-09-03 | 2006-03-09 | Biowisdom Limited | System and method for curating one or more multi-relational ontologies |
WO2006034038A2 (en) * | 2004-09-17 | 2006-03-30 | Become, Inc. | Systems and methods of retrieving topic specific information |
US7809763B2 (en) * | 2004-10-15 | 2010-10-05 | Oracle International Corporation | Method(s) for updating database object metadata |
US9137115B2 (en) * | 2004-12-06 | 2015-09-15 | Bmc Software, Inc. | System and method for resource reconciliation in an enterprise management system |
US20060167991A1 (en) | 2004-12-16 | 2006-07-27 | Heikes Brian D | Buddy list filtering |
US20060143227A1 (en) * | 2004-12-27 | 2006-06-29 | Helm Martin W | System and method for persisting software objects |
US8719779B2 (en) * | 2004-12-28 | 2014-05-06 | Sap Ag | Data object association based on graph theory techniques |
US7769579B2 (en) * | 2005-05-31 | 2010-08-03 | Google Inc. | Learning facts from semi-structured text |
US7685136B2 (en) * | 2005-01-12 | 2010-03-23 | International Business Machines Corporation | Method, system and program product for managing document summary information |
US20060238919A1 (en) | 2005-04-20 | 2006-10-26 | The Boeing Company | Adaptive data cleaning |
US20060248456A1 (en) | 2005-05-02 | 2006-11-02 | Ibm Corporation | Assigning a publication date for at least one electronic document |
US7590647B2 (en) | 2005-05-27 | 2009-09-15 | Rage Frameworks, Inc | Method for extracting, interpreting and standardizing tabular data from unstructured documents |
US7630977B2 (en) * | 2005-06-29 | 2009-12-08 | Xerox Corporation | Categorization including dependencies between different category systems |
US20070005593A1 (en) * | 2005-06-30 | 2007-01-04 | Microsoft Corporation | Attribute-based data retrieval and association |
US7797282B1 (en) | 2005-09-29 | 2010-09-14 | Hewlett-Packard Development Company, L.P. | System and method for modifying a training set |
US7493317B2 (en) * | 2005-10-20 | 2009-02-17 | Omniture, Inc. | Result-based triggering for presentation of online content |
US7730013B2 (en) * | 2005-10-25 | 2010-06-01 | International Business Machines Corporation | System and method for searching dates efficiently in a collection of web documents |
US7574449B2 (en) * | 2005-12-02 | 2009-08-11 | Microsoft Corporation | Content matching |
US8260785B2 (en) | 2006-02-17 | 2012-09-04 | Google Inc. | Automatic object reference identification and linking in a browseable fact repository |
US8954426B2 (en) | 2006-02-17 | 2015-02-10 | Google Inc. | Query language |
US7555471B2 (en) | 2006-01-27 | 2009-06-30 | Google Inc. | Data object visualization |
US7590628B2 (en) | 2006-03-31 | 2009-09-15 | Google, Inc. | Determining document subject by using title and anchor text of related documents |
US7610382B1 (en) | 2006-06-30 | 2009-10-27 | Amazon Technologies, Inc. | System and method for marking content |
US8458207B2 (en) * | 2006-09-15 | 2013-06-04 | Microsoft Corporation | Using anchor text to provide context |
US8719844B2 (en) | 2006-11-27 | 2014-05-06 | Morgan Stanley | Merging realtime data flows |
US8316007B2 (en) * | 2007-06-28 | 2012-11-20 | Oracle International Corporation | Automatically finding acronyms and synonyms in a corpus |
-
2006
- 2006-02-17 US US11/356,765 patent/US8244689B2/en active Active
-
2012
- 2012-08-14 US US13/585,775 patent/US20130041910A1/en not_active Abandoned
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020169730A1 (en) * | 2001-08-29 | 2002-11-14 | Emmanuel Lazaridis | Methods for classifying objects and identifying latent classes |
US20030177110A1 (en) * | 2002-03-15 | 2003-09-18 | Fujitsu Limited | Profile information recommendation method, program and apparatus |
US20060047838A1 (en) * | 2004-06-25 | 2006-03-02 | Abhishek Chauhan | Inferring server state in a stateless communication protocol |
Non-Patent Citations (3)
Title |
---|
Andritsos et al. "Information-Theoretic Tools for Mining Database Structure from Large Data Sets," ACM SIGMOD, 2004 * |
Chaudhuri et al. "Robust and Efficient Fuzzy Match for Online Data Cleaning," SIGMOD 2003 * |
Chen et al. "Exploiting Relationships for Object Consolidation," SIGMOD 2005 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2015183306A1 (en) * | 2014-05-30 | 2015-12-03 | Hewlett-Packard Development Company, L.P. | Processing a table of columns |
CN104899189A (en) * | 2015-05-27 | 2015-09-09 | 深圳市华傲数据技术有限公司 | Object name matching method based on information entropy |
Also Published As
Publication number | Publication date |
---|---|
US8244689B2 (en) | 2012-08-14 |
US20070198597A1 (en) | 2007-08-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8244689B2 (en) | Attribute entropy as a signal in object normalization | |
US10223406B2 (en) | Entity normalization via name normalization | |
US7672971B2 (en) | Modular architecture for entity normalization | |
US7966291B1 (en) | Fact-based object merging | |
US9558186B2 (en) | Unsupervised extraction of facts | |
US7774328B2 (en) | Browseable fact repository | |
US8260785B2 (en) | Automatic object reference identification and linking in a browseable fact repository | |
US8055674B2 (en) | Annotation framework | |
US7792837B1 (en) | Entity name recognition | |
US7464090B2 (en) | Object categorization for information extraction | |
US7555471B2 (en) | Data object visualization | |
US7454398B2 (en) | Support for object search | |
US8239751B1 (en) | Data from web documents in a spreadsheet | |
US8954426B2 (en) | Query language | |
US9208229B2 (en) | Anchor text summarization for corroboration | |
US7991797B2 (en) | ID persistence through normalization | |
US20070185870A1 (en) | Data object visualization using graphs | |
US20070185895A1 (en) | Data object visualization using maps | |
US20070179965A1 (en) | Designating data objects for analysis | |
US9785686B2 (en) | Corroborating facts in electronic documents | |
US7739212B1 (en) | System and method for updating facts in a fact repository |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO RESPOND TO AN OFFICE ACTION |