U.S. patent application number 14/875669 was filed with the patent office on 2016-02-04 for method and apparatus for providing internet service in mobile communication terminal.
The applicant listed for this patent is Samsung Electronics Co., Ltd.. Invention is credited to Jong-Woo Ha, Jeong-Soo Lee, Jung-Hyun Lee, Sang-Geun Lee, Kyu-Sun Shim.
Application Number | 20160036928 14/875669 |
Document ID | / |
Family ID | 45807698 |
Filed Date | 2016-02-04 |
United States Patent
Application |
20160036928 |
Kind Code |
A1 |
Lee; Jeong-Soo ; et
al. |
February 4, 2016 |
METHOD AND APPARATUS FOR PROVIDING INTERNET SERVICE IN MOBILE
COMMUNICATION TERMINAL
Abstract
A method and apparatus provide an Internet service in a mobile
communication terminal. The method includes determining a user
interest subject from user data existing within the mobile
communication terminal, collecting service items through network
access, determining a subject for each of the collected service
items, determining relevance between the user interest subject and
each of the service items, and recommending a service item
according to the relevance.
Inventors: |
Lee; Jeong-Soo;
(Gyeonggi-do, KR) ; Lee; Sang-Geun; (Seoul,
KR) ; Ha; Jong-Woo; (Seoul, KR) ; Lee;
Jung-Hyun; (Seoul, KR) ; Shim; Kyu-Sun;
(Seoul, KR) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Samsung Electronics Co., Ltd. |
Gyeonggi-do |
|
KR |
|
|
Family ID: |
45807698 |
Appl. No.: |
14/875669 |
Filed: |
October 5, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
13229022 |
Sep 9, 2011 |
9152723 |
|
|
14875669 |
|
|
|
|
Current U.S.
Class: |
709/203 |
Current CPC
Class: |
G06Q 30/02 20130101;
H04L 67/42 20130101; G06F 16/9535 20190101; H04L 67/16 20130101;
G06Q 10/107 20130101; H04L 67/22 20130101; H04W 8/183 20130101 |
International
Class: |
H04L 29/08 20060101
H04L029/08; H04L 29/06 20060101 H04L029/06; H04W 8/18 20060101
H04W008/18 |
Foreign Application Data
Date |
Code |
Application Number |
Sep 10, 2010 |
KR |
10-2010-0088709 |
Claims
1-20. (canceled)
21. A method in a mobile communication terminal, the method
comprising: implementing an application; registering at least one
user interest subject to a server by accessing a network;
collecting user data stored in the mobile communication terminal;
and performing at least one service based on the user interest
subject and the user data.
22. The method of claim 21, wherein the user data comprises at
least one of a short message, a multimedia message, an electronic
mail (e-mail), a file, a schedule, a memo, or Web-usage
information.
23. The method of claim 21, wherein the performing of the service
comprises: collecting at least one service item corresponding to
the user interest subject by accessing the network; and displaying
the service item.
27. An apparatus comprising: a display configured to display a
screen; and a controller configured to: implement an application,
register at least one user interest subject to a server by
accessing a network, collect user data stored in the apparatus; and
perform at least one service based on the user interest subject and
the user data.
28. The apparatus of claim 27, wherein the user data comprises at
least one of a short message, a multimedia message, an electronic
mail (e-mail), a file, a schedule, a memo, or Web-usage
information.
29. The apparatus of claim 27, wherein the controller is further
configured to: collect at least one service item corresponding to
the user interest subject by accessing the network, and display the
service item.
30. The apparatus of claim 27, wherein the controller is further
configured to: determine the user interest subject corresponding to
the application.
31. The apparatus of claim 27, wherein the controller is further
configured to: receive the service item from the server by
accessing the network.
32. The apparatus of claim 27, wherein the service item comprises
information on at least one of news, sports, shopping, or at least
another application.
Description
CROSS-REFERENCE TO RELATED APPLICATION(S) AND CLAIM OF PRIORITY
[0001] The present application is related to and claims priority
under 35 U.S.C. .sctn.119(a) to a Korean Patent Application filed
in the Korean Intellectual Property Office on Sep. 10, 2010 and
assigned Serial No. 10-2010-0088709, the contents of which are
herein incorporated by reference.
TECHNICAL FILED OF THE INVENTION
[0002] The present invention relates to a method and apparatus for
providing an Internet service in a mobile communication terminal.
More particularly, the present invention relates to a method and
apparatus for analyzing user data of a mobile communication
terminal and recommending a service in the mobile communication
terminal
BACKGROUND OF THE INVENTION
[0003] With the rapid development of mobile communication
technologies, various services using mobile communication terminals
are being provided. Particularly, in recent years, a service of
providing information on a user interest field considering the user
interest field in a mobile communication terminal, i.e., a
personalized information service of an on-demand scheme is being
provided.
[0004] FIG. 1 illustrates a method of providing a personalized
information service of an on-demand scheme according to the
conventional art.
[0005] As illustrated in FIG. 1, the conventional personalized
information service is of a scheme in which, if a mobile
communication terminal 100 registers a user interest field to a
user information storage server 102 (operation 110), the user
information storage server 102 provides the registered user
interest field to a personalized information provision server 104,
and the personalized information provision server 104 searches
service information associated with the user interest field and
then provides the searched service information to the mobile
communication terminal 100 (operation 112).
[0006] In the conventional personalized information service, a
mobile communication terminal should previously register a user
interest field to a server as above. That is, the conventional
personalized information service has a disadvantage that, unless
the mobile communication terminal previously registers the user
interest field to the server, the mobile communication terminal
cannot be provided with desired information. Further, in the
conventional personalized information service, whenever a user's
own interest field changes, a user has to register the changed
interest field to the server himself/herself, so there is a problem
of causing troublesomeness and inconvenience at user side.
SUMMARY OF THE INVENTION
[0007] To address the above-discussed deficiencies of the prior
art, it is a primary object to provide at least the advantages
below. Accordingly, one aspect of the present disclosure is to
provide a method and apparatus for analyzing user data and
recommending an Internet service in a mobile communication
terminal.
[0008] Another aspect of the present disclosure is to provide a
method and apparatus for determining a user interest subject from
user data, collecting Internet service items, determining a subject
of the service items, and then recommending a service item
corresponding to the user interest subject in a mobile
communication terminal.
[0009] A further aspect of the present disclosure is to provide a
method and apparatus for extracting term vectors from user data and
each service item and determining syntactic similarity between
respective term vectors in a mobile communication terminal
[0010] Yet another aspect of the present disclosure is to provide a
method and apparatus for extracting subjects from user data and
each service item and determining semantic similarity between
respective subjects in a mobile communication terminal
[0011] Still another aspect of the present disclosure is to provide
a method and apparatus for determining similarity between a user
interest subject and a service item subject, and recommending a
service according to the similarity in a mobile communication
terminal.
[0012] Still another aspect of the present disclosure is to provide
a method and apparatus for determining a term vector reflecting a
feature of a hierarchical structure between categories, for each
category of a subject classification tree in a mobile communication
terminal
[0013] Still another aspect of the present disclosure is to provide
a method and apparatus for determining relevance to other
categories for each category of a subject classification tree and,
based on the relevance, recommending a service corresponding to a
user interest subject in a mobile communication terminal.
[0014] The above aspects are achieved by providing a method and
apparatus for providing an Internet service in a mobile
communication terminal
[0015] According to one aspect of the present disclosure, a method
for providing an Internet service in a mobile communication
terminal is provided. The method includes determining a user
interest subject from user data existing within the mobile
communication terminal, collecting service items through network
access, determining a subject for each of the collected service
items, determining relevance between the user interest subject and
each of the service items, and recommending a service item
according to the relevance.
[0016] According to another aspect of the present disclosure, an
apparatus for providing an Internet service in a mobile
communication terminal is provided. The apparatus includes a user
interest subject determination unit, a service item collection and
classification unit, a service item ranking unit, and a service
recommendation unit. The user interest subject determination unit
determines a user interest subject from user data existing within
the mobile communication terminal. The service item collection and
classification unit collects service items through network access,
and determines a subject for each of the collected service items.
The service item ranking unit determines relevance between the user
interest subject and each of the service items. The service
recommendation unit recommends a service item according, to the
relevance.
[0017] Other aspects, advantages and salient features of the
disclosure will become apparent to those skilled in the art from
the following detailed description. Which, taken in conjunction
with the annexed drawings, discloses illustrative embodiments of
the disclosure.
[0018] Before undertaking the DETAILED DESCRIPTION OF THE INVENTION
below, k may be advantageous to set forth definitions of certain
words and phrases used throughout this patent document: the terms
"include" and "comprise," as well as derivatives thereof, mean
inclusion without limitation; the term "or," is inclusive, meaning
and/or; the phrases "associated with" and "associated therewith,"
as well as derivatives thereof, may mean to include, be included
within, interconnect with, contain, be contained within, connect to
or with, couple to or with, be communicable with, cooperate with,
interleave, juxtapose, be proximate to, be bound to or with, have,
have a property of, or the like; and the term "controller" means
any device, system or part thereof that controls at least one
operation, such a device may be implemented in hardware, firmware
or software, or some combination of at least two of the same. It
should be noted that the functionality associated with any
particular controller may be centralized or distributed, whether
locally or remotely. Definitions for certain words and phrases are
provided throughout this patent document, those of ordinary skill
in the art should understand that in many, if not most instances,
such definitions apply to prior, as well as future uses of such
defined words and phrases.
BRIEF DESCRIPTION OF THE DRAWINGS
[0019] For a more complete understanding of the present disclosure
and its advantages, reference is now made to the following
description taken in conjunction with the accompanying drawings, in
which like reference numerals represent like parts:
[0020] FIG. 1 is a diagram illustrating a method of providing a
personalized information service of an on-demand scheme according
to the conventional art;
[0021] FIG. 2 is a block diagram illustrating a construction of a
mobile communication terminal for providing a personalized
information service according to the present disclosure;
[0022] FIG. 3 is a block diagram illustrating a detailed
construction of a user interest subject determination unit in a
mobile communication terminal according to an embodiment of the
present disclosure;
[0023] FIG. 4 is a block diagram illustrating a detailed
construction of a service item collection and classification unit
in a mobile communication terminal according to an embodiment of
the present disclosure;
[0024] FIG. 5 is a block diagram illustrating a detailed
construction of a service item ranking unit in a mobile
communication terminal according to an embodiment of the present
disclosure;
[0025] FIG. 6 is a diagram illustrating user data in a mobile
communication terminal according to an embodiment of the present
disclosure;
[0026] FIG. 7 is a diagram illustrating an eXtensible Markup
Language (XML) file generated in a mobile communication terminal
according to an embodiment of the present disclosure;
[0027] FIG. 8 is a diagram illustrating a user interest subject
determined based on user data in a mobile communication terminal
according to an embodiment of the present disclosure;
[0028] FIG. 9 is a diagram illustrating an example of an Internet
service item collected in a mobile communication terminal according
to an embodiment of the present disclosure;
[0029] FIG. 10 is a diagram illustrating similarity between service
item subjects in a mobile communication terminal according to an
embodiment of the present disclosure;
[0030] FIG. 11 is a diagram illustrating a method for obtaining a
value when following the second circulation pattern in an
embodiment of the present disclosure;
[0031] FIG. 12 is a diagram illustrating a screen construction of
recommending a service item in a mobile communication terminal
according to an embodiment of the present disclosure; and
[0032] FIGS. 13A and 13B illustrate a procedure of recommending a
service item according to a user interest subject in a mobile
communication terminal according to an embodiment of the present
disclosure.
[0033] Throughout the drawings, like reference numerals will be
understood to refer to like parts, components and structures.
DETAILED DESCRIPTION OF THE INVENTION
[0034] FIGS. 2 to 13B, discussed below, and the various embodiments
used to describe the principles of the present disclosure in this
patent document are by way of illustration only and should not be
construed in any way to limit the scope of the disclosure. Those
skilled in the art will understand that the principles of the
present disclosure may be implemented in any suitably arranged
electronic device. Preferred embodiments of the present invention
will be described herein below with reference to the accompanying
drawings. In the following description, well-known functions or
constructions are not described in detail since they would obscure
the invention in unnecessary detail. And, terms described below,
which are defined considering functions in the present invention,
can be different depending on user and operator's intention or
practice. Therefore, the terms should be defined on the basis of
the disclosure throughout this specification
[0035] Below, illustrative embodiments of the present disclosure
provide a method and apparatus for analyzing user data and
recommending a service in a mobile communication terminal.
[0036] FIG. 2 illustrates a construction of a mobile communication
terminal for providing a personalized information service according
to the present disclosure.
[0037] Referring to FIG. 2, the mobile communication terminal 200
includes a user interest subject determination unit 210, a service
item collection and classification unit 220, a service item ranking
unit 230, and a personalized service recommendation unit 240.
[0038] The user interest subject determination unit 210 analyzes
user data existing within the mobile communication terminal and
determines a user interest subject. Here, the user data means data
such as a short message existing within the mobile communication
terminal, a multimedia message, an electronic-mail (e-mail), a
file, a schedule, a memo, Web-usage information and the like. In
detail, the user interest subject determination unit 210 extracts a
text from the user data existing within the mobile communication
terminal, analyzes the text, generates term vectors, classifies the
term vectors according to a subject classification tree embedded in
the mobile communication terminal, and determines a user interest
subject. Here, the subject classification tree is classifying, by
subject, concepts suitable for indicating user interest fields and
expressing the classified concepts in a tree structure. For
example, the subject classification tree can be an open directory
project widely known in the art. Undoubtedly, a general open
directory project is of as much wide range as being a Web
directory, so the present disclosure may extract certain categories
suitable for indicating user interest fields from the open
directory project, for use. The user interest subject determination
unit 210 is described later in detail with reference to FIG. 3.
[0039] The service item collection and classification unit 220
accesses the Internet 202, collects Internet service items,
analyzes texts of the collected service items, generates term
vectors, classifies the term vectors according to a subject
classification tree embedded in the mobile communication terminal,
and determines a subject of each of the collected service items.
The service item collection and classification unit 220 is
described later in detail with reference to FIG. 4.
[0040] The service item ranking unit 230 determines syntactic
similarity and semantic similarity using the term vector and user
interest subject for the user data and the term vector and service
item subject for the service items, determines the total similarity
between the user interest subject and the service item subject
using the syntactic similarity and semantic similarity, and
determines a relevance rank of each of the service items to the
user interest subject. The service item ranking unit 230 is
described later in detail with reference to FIG. 5.
[0041] The personalized service recommendation unit 240 controls a
function for, if a service item recommendation event occurs,
displaying a window for selecting the kind of recommendation
service item through a screen and, if the kind of recommendation
service item is selected, determining service items corresponding
to the selected kind in consideration of a relevance rank
determined in the service item ranking unit 230 and displaying a
list including the determined service items on the screen. Further,
the personalized service recommendation unit 240 controls and
processes a function for, if any one of the recommendation service
items is selected, displaying the detailed contents of the selected
service item on the screen. For example, as illustrated in FIG. 12,
the personalized service recommendation unit 240 controls and
processes a function for displaying a window for selecting which
kind of service item will be recommended among news or mobile
Applications (Apps), displaying a list including mobile App
services of a great relevance to a user interest subject and, if
the mobile App service is selected by a user, displaying detailed
information on the mobile App service selected by the user on the
screen.
[0042] Thus, a detailed construction of the mobile communication
terminal is described below with reference to FIGS. 3 to 5.
[0043] FIG. 3 illustrates a detailed construction of a user
interest subject determination unit in a mobile communication
terminal according to an embodiment of the present disclosure.
[0044] Referring to FIG. 3, the user interest subject determination
unit 210 includes a user data text extractor 310, a user data text
analyzer 320, and a user data term vector classifier 330.
[0045] The user data text extractor 310 extracts a text
representing a user interest subject from user data existing within
the mobile communication terminal For instance, the user data text
extractor 310 can extract a text from a short message, a multimedia
message, an e-mail, a file, a schedule, a memo, Web-usage
information and the like stored within the mobile communication
terminal as illustrated in FIG. 6. At this time, the user data text
extractor 310 extracts a generation time and generation position of
the extracted text, and metadata of a corresponding application
together, and stores the extraction result in an Extensible Markup
Language (XML) form. Here, the metadata of the application includes
at least one of the kind of the application, a feature, a revision
time, a generation time, and context information. Here, the context
information can be used for applying a weight at the time of
generating a term vector in the user data text analyzer 320. That
is, the user data text extractor 310 can extract a text from the
user data of FIG. 6 and, at this time, generate an XML file of FIG.
7.
[0046] The user data text analyzer 320 analyzes text data extracted
from the user data text extractor 310 and generates a term vector
according to a vector space model. Here, the term vector is
composed of individual terms existing in the text data, and can be
generated reflecting weights dependent on the importance of
respective terms within an extracted text. At this time, the
weights dependent on the importance of the respective terms can be
determined considering the frequency within the extracted text, and
a generation time and generation position of the text. For
instance, in an example where the frequency of appearance of a
specific term in texts provided from the user data text extractor
310 is high, the user data text analyzer 320 can determine the
specific term as a key term expressing a user interest subject and
set a high weight to the specific term. Further, the user data text
analyzer 320 can set weights to respective terms using a Term
Frequency Inverse Document Frequency (TFIDF) weight allocation
method widely known in the art, or can set weights using context
information recorded in an XML file generated in the user data text
extractor 310. For instance, the user data text analyzer 320 can
set higher weights to more recently generated terms based on the
context information recorded in the XML file and, through this, can
obtain an effect of being capable of reflecting a recent user
interest subject.
[0047] If term vectors for respective terms within a text are
generated in the user data text analyzer 320, the user data term
vector classifier 330 classifies the generated term vectors based
on a subject classification tree 340 embedded in the mobile
communication terminal and determines a user interest subject.
Here, the subject classification tree 340 classifies, by subject,
concepts suitable for indicating user interest fields and expresses
the classified concepts in a tree structure. For example, the
subject classification tree 340 can be an open directory project
known in the art. Undoubtedly, a general open directory project
known in the art is of as much wide range as being a Web directory,
so the present disclosure may extract certain some categories
suitable for indicating user interest fields from the open
directory project, for use. Here, each category of the subject
classification tree 340 can include a list of Web pages
corresponding to the each category. The list of Web pages may
include terms representing a characteristic of a corresponding
category.
[0048] The user data term vector classifier 330 can perform machine
learning for term vector classification with reference to the list
of Web pages included in each category of the subject
classification tree 340. At this time, a machine learning algorithm
can be Rocchio's algorithm, K-Nearest-Neighbor (KNN) algorithm,
Naive Bayes (NB) algorithm, Support Vector Machine (SVM) algorithm,
and the like widely known in the art, For instance, in a example
where Web pages `a`, and `c` are included in a category `A`, the
user data term vector classifier 330 may be learned to classify, as
the category `A`, term vectors corresponding to the Web page `a`.
After the learning is completed, if a user data term vector is
input, the user data term vector classifier 330 can determine a
category corresponding to the user data term vector with reference
to the subject classification tree 340, and determine a subject of
the determined category as a user interest subject corresponding to
the user data term vector. For instance, the user data term vector
classifier 330 may classify term vectors extracted from the user
data of FIG. 6 and, as illustrated in FIG. 8, classify the term
vectors into five categories and determine a subject of each
category as a user interest subject.
[0049] FIG. 4 illustrates a detailed construction of a service item
collection and classification unit in a mobile communication
terminal according to an embodiment of the present disclosure.
[0050] Referring to FIG, 4, the service item collection and
classification unit 220 includes a mobile Internet service item
collector 410, a service item text analyzer 420, and a service item
term vector classifier 430.
[0051] The mobile Internet service item collector 410 accesses the
mobile Internet 202 and collects service items (e.g., news and
mobile Apps) recommendable to a user. For instance, the mobile
Internet service item collector 410 collects the latest mobile App
information from a mobile App site suitable to an operation
environment of the mobile communication terminal, and collects the
latest news from a news portal site enabling information
collection. At this time, the mobile Internet service item
collector 410 can collect related service items using a user
interest subject determined in the user interest subject
determination unit 210.
[0052] The service item text analyzer 420 extracts a text from
service items collected in the mobile Internet service item
collector 410, analyzes the extracted text, and generates a term
vector according to a vector space model. Here, the term vector is
composed of individual terms existing in the text of the collected
service items, and reflects weights dependent on the importance of
respective terms within the extracted text. Here, the weights
dependent on the importance of respective terms can be determined
considering the frequency of each term within the extracted text
and a generation time and generation position of the text. That is,
the service item text analyzer 420 may set the weights dependent on
the importance of the respective terms in the same method as that
of the user data text analyzer 320.
[0053] The service item term vector classifier 430 classifies term
vectors generated in the service item text analyzer 420 based on a
subject classification tree 44l embedded in the mobile
communication terminal, and determines a subject of each of the
collected service items. Here, the service item term vector
classifier 430 classifies the term vectors in the same method as
that of the user data term vector classifier 330 and determines a
corresponding subject. Further, the subject classification tree 440
referred in the service item term vector classifier 430 is the same
as the subject classification tree 340 referred in the user data
term vector classifier 330.
[0054] FIG. 5 illustrates a detailed construction of a service item
ranking unit in a mobile communication terminal according to an
embodiment of the present disclosure.
[0055] Referring to FIG. 5, the service item ranking unit 230
includes a syntactic matching unit 510, a semantic matching unit
520, and an integration ranking unit 530.
[0056] The syntactic matching unit 510 determines syntactic
similarity between a term vector generated in the user data text
analyzer 320 and a term vector generated in the service item text
analyzer 420. By determining the cosine similarity of a vector
space model according to Equation 1 below, the syntactic matching
unit 510 determines syntactic similarity between a term vector for
user data and a term vector for service items.
[0057] Equation 1 below represents a formula of determining cosine
similarity.
SyntacticScore ( u .fwdarw. , s i .fwdarw. ) = cos ( u .fwdarw. , s
i .fwdarw. ) = u .fwdarw. s i .fwdarw. u .fwdarw. s i .fwdarw. ( 1
) ##EQU00001##
[0058] Here, the `{right arrow over (u)}` represents a term vector
for user data, and the `{right arrow over (s.sub.i)}` represents a
term vector for service items.
[0059] The semantic matching unit 520 determines semantic
similarity between a user interest subject extracted from the user
data term vector classifier 330 and a service item subject
determined in the service item term vector classifier 430. By
applying a weighted Personalized PageRank (wPPR) algorithm to a
similarity graph 540 representing semantic similarity between
respective categories in a subject classification tree, the
semantic matching unit 520 determines the semantic similarity
between the user interest subject and the service item subject.
Here, the similarity graph 540 is a conversion of a hierarchical
tree structure of respective categories into a graph structure
connected according to semantic similarity between categories in
the subject classification tree. Each node of the similarity graph
540 represents each category of the subject classification tree,
and a link between respective nodes represents the existence of
semantic similarity between corresponding categories. Further, the
wPPR algorithm, which is the application of a weight to a
Personalized PageRank algorithm widely known in the art, is
described below in detail.
[0060] A method for generating links between respective nodes of
the similarity graph 540 is of three operations as follows. The
similarity graph 540 may be generated in the semantic matching unit
520, or may be generated in a different function block of the
mobile communication terminal.
[0061] Operation 1 is the operation of determining a centroid
vector for each category of the subject classification tree
according to Equation 2 below. The centroid vector, a vector being
representative of learning data of each category, is an average
term vector of the learning data. Here, the learning data can be
Web pages of an open directory project used at the time of machine
learning of the user data term vector classifier 330 or service
item term vector classifier 430.
[0062] Equation 2 below represents a formula of determining a
centroid vector of each category.
u .fwdarw. ( c ) = 1 D c d .di-elect cons. D c v .fwdarw. ( d ) ( 2
) ##EQU00002##
[0063] The means a category, the `{right arrow over (.mu.)}(c)`
means a centroid vector of the category `c`, the `D.sub.c` means a
learning data set of the category `c` and the `{right arrow over
(v)}(d)` a means a term vector for learning data `d.`
[0064] Operation 2 is the operation of determining a merge centroid
vector for each category of the subject classification tree
according to Equation 3 below. The merge centroid vector represents
the reflection of a feature of a hierarchical structure of the
subject classification tree in the centroid vectors of the
respective categories. That is, the centroid vector includes only a
feature of a corresponding category and does not reflect
hierarchical relationship between categories within the subject
classification tree, but the merge centroid vector represents the
inclusion of features of centroid vectors of descendent eatery a
centroid vector of a parent category such that the parent category
can include features of child categories.
[0065] Equation 3 below represents a formula of determining a merge
centroid vector.
u .fwdarw. ' ( c ) = 1 1 + child ( c ) ( .mu. .fwdarw. ( c ) .mu.
.fwdarw. ( c ) + c k .di-elect cons. child ( c ) .mu. .fwdarw. ' (
c k ) .mu. .fwdarw. ' ( c k ) ) ( 3 ) ##EQU00003##
[0066] Here, the `{right arrow over (.mu.)}(c)` means a merge
centroid vector or a category `c`, the `child(c)` means the set of
child categories of the category `c`, c.sub.k means a k.sup.th
categ, and the `{right arrow over (.mu.)}(c)` means a centroid
vector for the category `c`.
[0067] Operation 3 is the operation of determining semantic
similarity between all categories. The semantic similarity between
the categories means cosine similarity between merge centroid
vectors for the categories. Here, the semantic matching unit 520
compares the semantic similarity between the categories with a
threshold value. In an example where the semantic similarity
between the categories is greater than the threshold value, the
semantic matching unit 520 generates a link between corresponding
categories, generating a similarity graph. At this time, as
illustrated in FIG. 10, the semantic matching unit 520 sets the
semantic similarity as a weight for the generated link. Here,
because the weight for the link is determined by the cosine
similarity between the merge centroid vectors, the weight for the
link can represent a feature of a hierarchical structure of the
subject classification tree.
[0068] The semantic matching unit 520 ranks semantic similarities
of other categories for each category in the generated similarity
graph 540 according to the wPPR algorithm proposed in the present
disclosure.
[0069] For the sake of this, first, based on a link weight of the
similarity graph 540, the semantic matching unit 520 determines a
relevance matrix (R). In the relevance matrix (R), a (i, j)
component (r.sub.ij) means semantic similarity of an i.sup.th
category for a j.sup.th category. That is, the semantic matching
unit 520 determines a probability that a random surfer circulating
a similarity graph makes a visit to each category, using a
personalized PageRank that is one of Markov Random Walk Models
widely known in the art. In a little detail, the semantic matching
unit 520 can determine a probability that the random surfer makes a
visit, to the i.sup.th category from the j.sup.th category,
determine the determined value as semantic similarity between the
two categories, and rank semantic similarities of other categories
for the j.sup.th category according to a size of the semantic
similarity.
[0070] Here, a pattern in which the random surfer circulates the
similarity graph can be defined as two examples. According to the
first circulation pattern, the random surfer circulates the
similarity graph at a probability of `(1-d)` every moment and,
according to the second circulation pattern, circulates the
similarity graph at a probability of `d.` Here, the `d` is a
damping factor, and can have a real number of `0` to `1`. According
to experiments, an optimal value of the `d` can be found
empirically. In an example where the random surfer follows the
first circulation pattern, the random surfer makes a visit to a
category reliable within the similarity graph, i.e., a j.sup.th
category being currently in visit in the present disclosure. In an
example where the random surfer follows the second circulation
pattern, the random surfer makes a visit to a category linked with
a category being currently in visit, at a probability proportional
to a link weight. A probability that the random surfer moves to a
next category when following the second circulation pattern is
determined according to Equation 4 below.
[0071] Equation 4 below is a formula for determining a probability
of movement of a random surfer.
tw ij = sim ( c i , c j ) c k .di-elect cons. N ( C j ) sim ( c k ,
c j ) ( 4 ) ##EQU00004##
[0072] Here, the `tw.sub.ij` represents a probability that the
random surfer moves from a j.sup.th category to an i.sup.th
category, the `sim(c.sub.i, c.sub.i)` represents similarity between
the categories, i.e., a link weight between the categories, and the
`N(c.sub.j)` represents the set of categories connecting with
c.sub.j.
[0073] FIG. 11 illustrates a method for obtaining a `tw.sub.ij`
value when following the second circulation pattern in an
illustrative embodiment of the present disclosure.
[0074] Referring to FIG. 11, a probability (i.e., a `tw.sub.ij`
value) of movement between categories within a subject
classification tree is determined based on a link weight between
respective categories illustrated FIG. 10. For instance, assuming
that a random surfer currently makes a visit to a `Sports` category
1001 and circulates a similarity graph according to the second
circulation pattern at a probability of `d`, of probability of
movement to each of `Soccer`, `Baseball`, and `Shopping` categories
1003, 1005, and 1007 linked to the `Sports` category 1001 is
determined according to a ratio of sum (1.7=0.7 (1013)+0.7
(1015)+0.3 (1017)) of weights of the total links connected with the
`Sports` category 1001 to weight of a corresponding link. That is,
as illustrated in FIG. 11, a probability that the random surfer
moves from the `Sports` category 1001 to the `Soccer` category 1003
is equal to `0.7/1.7` (1113), and a probability that the random
surfer moves from the `Sports` category 1001 to the `Baseball`
category 1005 is equal to `0.7/1.7` (1115), and a probability that
the random surfer moves from the `Sports` category 1001 to the
`Shopping` category 1007 is equal to `0.3/1.7` (1117).
[0075] Based on the definition of the two circulation patterns of
the random surfer, the `r.sub.ij` can be determined according to
Equation 5 below.
r ij = d [ c k .di-elect cons. I ( c i ) tw ik r kj ] + ( 1 - d ) t
ij ( 5 ) ##EQU00005##
[0076] Here, the `I(c.sub.i)` represents the set of categories
having a link to `c.sub.i`, and the `t.sub.ij` is for determining
the first circulation pattern. In an example where a current
category is set to `c.sub.j`, the represents a trusted weight of
the `c.sub.i`, in an example where `i` is equal to `j`, the
`i.sub.ij` is set to `1 ` and, in remnant examples, the `t.sub.ij`
is set to `0`.
[0077] The definition of Equation 5 above using a matrix notation
method can be expressed according to Equation 6 below.
R.sub.1=d[WR.sub.i-1]+(1-d)T (6)
[0078] Here, the `R` represents a relevance matrix determined
according to a wPPR algorithm, and the `W` represents a transition
matrix and has the same (i,j).sup.th component as the `tw.sub.ij`
of Equation 4 above. The `T`, a trusted matrix, has the same (i,
j).sup.th component as the `t.sub.ij` of Equation 5 above, so the
`T` becomes a unit matrix.
[0079] That is, the semantic matching unit 520 can digitize
semantic similarity between arbitrary categories on the basis of
the relevance matrix of Equation 6 above. At this time, the
category can be a category corresponding to a user interest subject
or service item subject. Accordingly, in a example where the user
interest subject is determined as a category (c.sub.j), the
semantic matching unit 520 can determine, as a (i, j) component
value of the relevance matrix, semantic similarity between the
category (c.sub.i) corresponding to the service item subject and
the category (c.sub.j),
[0080] By linearly combining syntactic similarity and semantic
similarity, determined in the syntactic matching unit 510 and the
semantic matching unit 520 respectively, the integration ranking
unit 530 determines the total similarity according to Equation 7
below,
[0081] Equation 7 below represents a formula of determining the
total similarity.
TotalScore(u,s.sub.i)=(1-.lamda.).times.SyntacticScore({right arrow
over (u)},{right arrow over
(s.sub.i)})+.lamda..times.SemanticScore(uc,sc) (7)
[0082] Here, the `uc` is a category corresponding to a user
interest subject extracted from user data, and the `sc.sub.i`
represents a category corresponding to a service item subject. The
`.lamda.`, a weight for semantic similarity in a linear
combination, has a value of `0` to `1.0`, and can be determined
through experiment.
[0083] The integration ranking unit 530 determines relevance ranks
of service items to be recommended to a user based on the total
similarity determined through Equation 7 above.
[0084] FIGS. 13A and 13B illustrate a procedure of recommending a
service item according to a user interest subject in a mobile
communication terminal according to an illustrative embodiment of
the present disclosure.
[0085] Referring to FIG. 13A and 13B, in operation 1301, the mobile
communication terminal determines whether a preset user data
collection period is present. In an example where the user data
collection period is present, the mobile communication terminal
proceeds to operation 1303 and extracts a text from user data
existing within the mobile communication terminal. Here, the mobile
communication terminal can extract the text from the user data such
as a short message, a multimedia message, an e-mail, a file, a
schedule, a memo, Web-usage information and the like. After that,
in operation 1305, the mobile communication terminal generates term
vectors based on the extracted text of the user data. And then, in
operation 1307, the mobile communication terminal classifies the
term vectors according to a subject classification tree embedded in
the mobile communication terminal and determines a user interest
subject and then, proceeds to operation 1317 below.
[0086] In contrast, in an example where the user data collection
period is not present, the mobile communication terminal proceeds
to operation 1309 and determines whether a preset service item
collection period is present. In an example where the service item
collection period is not present, the mobile communication terminal
returns to operation 1301 and again performs the subsequent
operations. In contrast, in an example where the service item
collection period is present, the mobile communication terminal
proceeds to operation 1311 and accesses the mobile Internet,
collects Internet service items, and extracts a text from the
collected service items. After that, the mobile communication
terminal proceeds to operation 1313 and generates term vectors
based on the extracted text of the service items and, in operation
1315, classifies the term vectors according to the subject
classification tree and determines a service item subject and then,
proceeds to operation 1317 below.
[0087] In operation 1317, the mobile communication terminal
determines if a service item recommendation event takes place by a
user. When the service item recommendation event does not occur,
the mobile communication terminal returns to operation 1301 and
again performs the subsequent operations. In contrast, when the
service item recommendation event occurs, the mobile communication
terminal proceeds to operation 1319 and determines syntactic
similarity between a term vector for user data and a term vector
for service items. At this time, the mobile communication terminal
can determine the syntactic similarity between the user data term
vector and the service item term vector, using Equation 1
above.
[0088] After that, in operation 1321, the mobile communication
terminal determines semantic similarity between the user interest
subject and the service item subject. Here, the mobile
communication terminal connects respective categories in a subject
classification tree according to the semantic similarity,
determines a weight for a link between the respective categories,
determines a probability of movement of a random surfer on the
basis of a general PageRank algorithm widely known in the art, and
determines semantic similarity between the respective categories,
thereby being capable of determining the semantic similarity
between the user interest subject and the service item subject.
That is, the mobile communication terminal can determine the
semantic similarity between the user interest subject and the
service item subject using Equation 5 above.
[0089] Next, in operation 1323, the mobile communication terminal
determines a relevance rank representing the total similarity
between the user interest subject and the service item subject,
based on the determined syntactic similarity and semantic
similarity. After that, in operation 1325, the mobile communication
terminal recommends a service item according to the determined
relevance rank. Here, the mobile communication terminal can
determine the total similarity between the user interest subject
and the service item subject using Equation 7 above, and can
determine to have a higher relevance rank as higher is the total
similarity value of the service item subject for the user interest
subject.
[0090] After that, the mobile communication terminal terminates the
algorithm according to the present disclosure.
[0091] Here, a description is made for collecting user data and
service items every constant period but, only in an example where a
service item recommendation event takes place by a user, the mobile
communication terminal may collect the user data and the service
items.
[0092] As described above, illustrative embodiments of the present
disclosure have an effect of, by determining a user interest
subject from user data, collecting Internet service items,
determining a subject of the service items, determining similarity
between the user interest subject and the service item subject, and
recommending a service according to the similarity, being capable
of, even without separate input of the user interest subject,
analyzing the user interest subject based on data within a mobile
communication terminal and recommending a related Internet service
item in the mobile communication terminal. Further, the
illustrative embodiments of the present disclosure have an effect
of, because analyzing the user interest subject based on the data
within the mobile communication terminal, being capable of
recommending a suitable Internet service item properly
corresponding to the user interest subject changing every hour.
Further, the illustrative embodiments of the present disclosure
have an effect of, instead of transmitting user person's
information to the external through a network or storing the user
person's information in a server, analyzing the user interest
subject based on the data within the mobile communication terminal
and recommending a suitable Internet service item, thereby being
capable of protecting user person's data in the mobile
communication terminal
[0093] While the invention has been shown and described with
reference to certain preferred embodiments thereof, it will be
understood by those skilled in the art that various changes in form
and details may be made therein without departing from the spirit
and scope of the invention as defined by the appended claims.
* * * * *