A method and a computing device in which a meaning definition is configured as a semantic meta-set

Abstract

To provide an improved method for representing semantic attributes and semantic relations as semantic definitions on a computing device, and a non-transitory computer readable medium.SOLUTION: The method for representing semantic definitions includes a step 110 of constructing a semantic definition text, a step 120 of storing the semantic definition text in a metaset, a step 130 of converting the metaset to a digital data structure, and a step 140 of storing the digital data structure in a memory storage device of a computing device.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to the field of meaning definition. More specifically, the present invention relates to representing meaning definition in a computing device as semantic metadata for use in semantic processing and configurable semantic search.

Background Art

[0002] An object can be described by its visual and non-visual semantic attributes, and based on the definitions provided to a database during the steps of visual processing, as described in Japanese Patent No. 6830514, the relationships between semantic attributes are determined. For example, when an object describes a lemon, the database can provide the semantic relationships between "lemon" and semantic attributes such as "citrus fruit", "sour taste", "vitamin C", and "edible".

[0003] Considering the extremely large number of semantic attributes and the complexity of the various types of relationships between semantic attributes that can exist over just a small set of objects, there are many challenges in representing these semantic attributes and semantic relationships as meaning definitions on a computing device. Therefore, in this technical field, improved methods and data structures for representing meaning definitions are needed. Furthermore, there is a need in this technical field for the means to efficiently create, edit, search, and query databases of meaning definitions.

[0004] Description of Related Art Japanese Patent No. 6830514

Summary of the Invention

[0005] This application discloses a method for representing semantic definitions on a computing device. The semantic definition text is composed using operators. The semantic definition text includes a semantic concept text using a semantic concept operator and a semantic context text using a semantic context operator. The semantic definition text is stored in a metaset. The metaset is converted into a digital data structure and stored in the memory storage device of the computing device.

[0006] The semantic definition text may also include a semantic label text, a semantic label collection text, a semantic domain text, a semantic inheritance text, a semantic instance text, a vocabulary function text, a semantic internationalization text, or an operator for semantic reference.

[0007] This application further discloses a method for semantically searching for visuals. The input is received via a computer interface, and the semantic representation is derived from the input. The type of semantic definition is determined to be applied to the search. The type of semantic definition may include a semantic element definition, a semantic concept definition, a semantic context definition, a semantic label definition, a semantic domain definition, a semantic instance definition, a semantic inheritance definition, a semantic internationalization definition, a vocabulary function definition, or a semantic reference definition. A metaset having a semantic definition text representing the semantic definition can also be applied to the search. The search is executed based on the input, the determined type of semantic definition, and the selected metaset. A list of visuals determined to be semantically relevant to the input based on the search is output.

[0008] This application further discloses a method for selecting a semantic definition from a metaset. The semantic definition text representing the semantic definition is retrieved from the metaset. Each semantic definition text includes a subject operand, an operator, and at least one object operand. A set of semantic representations based on the semantic definition text is output. The semantic representations have selectable items corresponding to the operands. A selection of a selectable item is received. The selected semantic definition is determined based on the set of semantic representations and the selection of the selectable item. The selection of the semantic definition from the metaset can be used for metaset navigation or metaset filtering.

[0009] The configuration, editing, querying, saving, searching, or other interactions with the metaset can be performed using a semantic definition management editor.

[0010] The method disclosed in this application can be stored in a non-transitory computer-readable medium as instructions executable by a processor.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Modes for Carrying Out the Invention

[0012] In the following detailed description, reference is made to the accompanying drawings, which form a part hereof. In the drawings, like symbols generally identify like components unless the context dictates otherwise. The illustrative embodiments described in the detailed description, the drawings, and the claims are not intended to be limiting. Other embodiments may be utilized and other changes may be made without departing from the spirit or scope of the subject matter presented herein. Aspects of the present disclosure can be arranged, substituted, combined, separated, and designed in a variety of different configurations, as generally described herein and shown in the drawings, all of which are expressly contemplated herein and will be readily understood.

[0013] FIG. 1 shows a method of representing a semantic definition on a computing device 100 according to an embodiment of the present invention. Representing a semantic definition according to a method of representing a semantic definition on a computing device 100 has several technical advantages over the prior art. A metaset provides an efficient data structure for representing semantic attributes and relationships between various semantic attributes. The metaset is easily storable and accessible for various functions such as visual definition processing, editing, querying, filtering, navigation, and searching. The representation of the semantic definition as a semantic definition sentence can be configured to strictly follow the exact semantic definition described in Japanese Patent No. 6830514 and implement it. The design and implementation of the metaset and the semantic definition sentence are flexible (the semantic definition can be represented across an unlimited knowledge area), extensible (the various semantic definitions of Japanese Patent No. 6830514 can be implemented with new definitions such as semantic reference, semantic internationalization, and even unlimited unspecified definitions), and personalized / specialized / democratized (knowledge can be represented by experts / individuals with different semantic definitions via a metaset having references to knowledge sources).

[0014] In step 110, a semantic definition sentence is constructed. The semantic definition sentence represents a semantic definition that describes semantic attributes and their relationships. The semantic definition sentence can represent various types of semantic definitions described in Japanese Patent No. 6830514, including, but not limited to, semantic elements, semantic concepts, semantic contexts, semantic markers, semantic domains, semantic inheritance, semantic instances, and vocabulary functions. The semantic definition sentence can further represent semantic internationalization and semantic reference (described later). The semantic definition sentence can further represent an unspecified semantic definition. The semantic definition can be configured with a semantic definition management editor as shown in FIG. 2. Alternatively, the semantic definition may be configured with any program capable of text editing.

[0015] The semantic definition sentence is constructed using operators and operands. The semantic definition sentence generally includes a subject operand, an operator, and an object operand. Some semantic definitions may include multiple subject operands, multiple operators, or multiple object operands. The subject operand, operator, and object operand may appear in any order. In one embodiment, in order from left to right, there are a subject operand, an operator, and an object operand. The operand may be a word, and the operator may be a symbol and / or alphanumeric. The semantic definition sentence can be configured in any language or character set, including a two-byte language such as Japanese. Each semantic definition sentence may end with a semicolon or another character to indicate the end of the sentence.

[0016] A semantic element is the most basic component of a semantic definition sentence. The semantic element can be a visual or non-visual semantic attribute. The semantic element sentence may include a semantic element operator indicating that a block of text is defined as a semantic element. In an exemplary embodiment, two angle brackets are semantic element operators that define the text within the brackets, such as "[grape]", as a semantic element. The semantic element may generally refer to individual elements of the semantic definition sentence, such as semantic concepts, semantic domains, semantic instances, etc.

[0017] A semantic concept is a general concept, classification, or hierarchical structure that represents semantic relationships. A semantic concept sentence expresses a semantic concept relationship through one or more semantic elements that represent the semantic concept as a subject operand, a semantic concept operator, and an object operand. In an exemplary embodiment, a general semantic concept sentence can have the form “[A] :: [B], [C], [D];” where “::” is a general semantic concept operator and the object operands are separated by commas. For example, “[Berry] :: [Grape], [Raspberry], [Blueberry]; is a general semantic concept sentence that defines “Berry” as a semantic concept having semantic element members “Grape,” “Raspberry,” and “Blueberry.”

[0018] To define certain types of semantic concept relationships, other semantic concept operators may be used instead of the general semantic concept operator. Semantic concept relationships in a hierarchical structure that describe a “type of” relationship, an “is a” relationship, or a taxonomic structure are defined as hypernym / hyponym relationships. In an exemplary embodiment, “::>” is a hypernym operator and “<::” is a hyponym operator. For example, “[Fruit] ::> [Grape], [Orange], [Mango]; may be a hypernym semantic concept sentence, and “[Grape], [Raspberry], [Blueberry] <:: [Berry]; may be a hyponym semantic concept sentence.

[0019] Semantic concept relationships in a hierarchical structure that describe a whole / part relationship are defined as holonym / meronym relationships. In an exemplary embodiment, “ト” is a holonym operator and “ヨ” is a meronym operator. For example, “[Fruit] ト [Skin], [Flesh], [Seed]; is a holonym semantic concept sentence, and “[Root], [Trunk], [Branch], [Bark], [Leaf], [Flower] ヨ [Tree]; is a meronym semantic concept sentence.

[0020] The semantic concept relationship in which an exact action word is used to define a general verb is defined as a troponym relationship. In an exemplary embodiment, " / / >" and " / / <" are troponym operators. For example, "[Cut] / / > [Cut off], [Cut thinly];" "[Walk] / / > [Take a walk];" "[Snack], [Gnaw] / / < [Eat];" are troponym semantic concept sentences.

[0021] Certain types of semantic concept relationships not indicated by pre-defined operators may be indicated using free semantic concept operators. In an exemplary embodiment, the free semantic concept operator may be a pair of double brackets added to a semantic concept sentence using a general semantic concept operator, and the double brackets may enclose a free text description of the semantic concept relationship. For example, "[Antioxidant] :: (( Protects cell membranes from radical damage ));" is a free semantic concept sentence defining the semantic concept "antioxidant" as having a semantic concept relationship of the free text "Protects cell membranes from radical damage" because antioxidants protect cell membranes from radical damage. In another example, "[Antioxidant] :: [Vitamin C], [Manganese], (( Protects cell membranes from radical damage ));" is a free semantic concept sentence having a semantic concept relationship of "antioxidant" because "Vitamin C" and "Manganese" are antioxidants that protect cell membranes from radical damage.

[0022] A semantic context defines semantic relationships between two or more semantic concepts. A semantic context sentence expresses a semantic context relationship through at least a first semantic concept as a subject operand, a semantic context operator, and a second semantic concept as an object operand. The semantic context relationship can be further defined using another semantic context operator and one or more semantic elements. In an exemplary embodiment, a semantic context sentence may have the form “[A] μ [B];” where A and B are semantic concepts and “μ” is a semantic context operator. For example, “[Plant] μ [Soil] ;” is a semantic context sentence that defines the semantic context relationship between “Plant” and “Soil”. Another semantic context sentence may have the form “[A] μ [B] :: [C];” where “::” is a second semantic context operator and C is a semantic element that describes the semantic context relationship. The second semantic context operator may be the same symbol as a general semantic concept operator. For example, “[Plant] μ [Soil] :: [Nutrient];” is a semantic context sentence that defines the semantic context relationship between “Plant” and “Soil”. In other words, “Plant” and “Soil” are semantically contextually related through “Nutrient”.

[0023] Certain types of semantic context relationships not indicated by a predefined semantic context operator may be indicated using a free semantic context operator. In an exemplary embodiment, the free semantic context operator may be a set of double brackets added to a semantic context sentence, and the double brackets may enclose a free text description of the semantic context relationship. For example, “[Anti - inflammation] μ [Grape] :: (( * Research conducted with grape extract ));” describes a free semantic context where the relationship between anti - inflammation and grape is supported by research conducted with grape extract.

[0024] Semantic labels define semantic relationships between semantic elements, such as semantic concepts and semantic context relationships, in specific terms. A semantic label statement may have the same syntax as other types of statements, such as semantic concept statements or semantic context statements, but may also have semantic label operators that replace other operators. The text associated with the semantic label operator may represent a semantic label. In an exemplary embodiment, the semantic label operator is a set of single parentheses that enclose the descriptive text of the semantic label. In this exemplary embodiment, a semantic concept statement having the syntax "[A] :: [B], [C];" or a semantic context statement having the syntax "[A] ⋅ [B], [C];" may be further described by a corresponding semantic label statement having the syntax "[A] (semantic label) [B], [C];" For example, the semantic concept statement "[grape] :: [raisin], [wine];" may be further defined using the semantic label statement "[grape] (processed to become...) [raisin], [wine];" Similarly, the semantic context statement "[plant] ⋅ [soil];" may be further defined using the semantic label statement "[plant] (grown in...) [soil];"

[0025] A plurality of semantic labels related to a common operand may be collected as a semantic label collection. For example, the common operand may be a semantic concept that appears as a subject operand in a plurality of semantic label texts having different semantic labels. A semantic label collection text represents a semantic label collection relationship using a semantic label collection operator. In an exemplary embodiment, the semantic label collection text may have the form "[A] ha (semantic label 1), (semantic label 2);", where "ha" is a semantic label collection operator. For example, the semantic concept "[Fruit]" may appear in the following semantic concept text and semantic context text "[Fruit] :: [Juice];", "[Fruit] :: [Grape];", "[Fruit] :: [Citrus Fruit];", "[Fruit] mi [Orchard];", and "[Fruit] mi [Health]". These semantic concept texts and semantic context texts may be further described in the following semantic label texts, "[Fruit] (processed to become...) [Juice];", "[Fruit] (type) [Grape];", "[Fruit] (diversity) [Citrus Fruit];", "[Fruit] (...grown in) [Orchard];", and "[Fruit] (suitable for) [Health]". The semantic labels used in these semantic label texts are associated by sharing the semantic concept "[Fruit]", but may be represented as a semantic label collection in the semantic label collection text, "[Fruit] ha (...grown in), (processed to become...), (type), (diversity), (suitable for);".

[0026] The semantic domain defines a domain of knowledge that intersects with many other domains of knowledge. The semantic domain may be less specific than semantic concepts. The semantic domain relationship may be different from the semantic concept relationship in that the semantic domain relationship is not constrained by structural or hierarchical properties. A semantic domain statement may include a semantic domain, a semantic domain operator, and semantic elements, semantic concepts, semantic markers, or another semantic domain assigned to the semantic domain. In an exemplary embodiment, a semantic domain statement has the general form "[A] le <<semantic domain>>;", where "le" is a semantic domain operator and the operator "<< >>" indicates the semantic domain. For example, the semantic domain statement "[Cardiovascular], [Respiratory] le <<Medicine>>;" assigns the semantic elements or semantic concepts "Cardiovascular" and "Respiratory" to the semantic domain of "Medicine". The semantic domain statement "(grown at...) le <<Topography>>;" assigns the semantic marker "...grown at" to the semantic domain of "Topography". For example, multiple semantic markers can be assigned to one semantic domain in one semantic domain statement, such as "(suitable for), (bringing benefits) le <<Health>>;".

[0027] One semantic domain can be assigned to another semantic domain, such as "<City>> le <<Topography>>;". The semantic domain may be a trans-semantic domain that is one of two mutually exclusive semantic domains. For example, a pair of trans-semantic domains can be "<<Edible>>" and "<<Inedible>>". The semantic domain statement "<Material>> le <<Edible>>;" assigns the semantic domain "Material" to the trans-semantic domain "Edible". In this case, <<Material>> may not be assigned to <<Inedible>>.

[0028] By semantic inheritance, the semantic relationships of semantic elements or semantic concepts defined through semantic concepts or semantic contexts can be extended to other semantic elements or semantic concepts. Semantic inheritance can be applied to siblings of semantic elements or semantic concepts, to children of semantic elements or semantic concepts (downward), or to parents of semantic elements or semantic concepts (upward), for a specified number of levels (N levels), or for all levels of semantic concepts or semantic contexts of semantic elements or semantic concepts marked for inheritance.

[0029] In an exemplary embodiment, a semantic inheritance statement for siblings applied to a semantic concept statement has the form "ts [A] :: [X], [Y], [Z];", where "ts" is a semantic inheritance operator indicating semantic inheritance at the sibling level applied to the semantic concept statement "[A] :: [X], [Y], [Z];". For example, if there is a semantic concept statement "[Fruit] :: [Grape], [Tomato], [Persimmon];", the semantic inheritance statement "ts [Grape] :: [Antibacterial agent];" causes the semantic concept relationships with "[Antibacterial agent]" ("[Tomato] :: [Antibacterial agent];" and "[Persimmon] :: [Antibacterial agent];") to be inherited by "Tomato" and "Persimmon", which are siblings of "Grape" in the above semantic concept statement.

[0030] In an exemplary embodiment, a semantic inheritance statement for siblings applied to a semantic concept statement has the form "ts [A] m [B] :: [X], [Y], [Z];", where "ts" is a semantic inheritance operator indicating semantic inheritance at the sibling level applied to the semantic context statement "[A] m [B] :: [X], [Y], [Z];". For example, if there is a semantic concept statement "[Fruit] :: [Grape], [Tomato], [Persimmon];", the semantic inheritance statement "ts [Grape] m [Health] :: [Antibacterial agent], [Anti-inflammatory];" inherits the semantic concept relationship with "Health" described by "[Antibacterial agent], [Anti-inflammatory]" ("[Tomato] m [Health] :: [Antibacterial agent], [Anti-inflammatory];" and "[Persimmon] m [Health] :: [Antibacterial agent], [Anti-inflammatory];") for "Tomato" and "Persimmon", which are siblings of "Grape" in the above semantic concept statement.

[0031] The semantic concept relationship can be extended N levels downward to the children of the semantic element or semantic concept. This can be done by indicating the downward direction and the inheritance level by the semantic inheritance operator and the intended semantic element or semantic concept. A computer system that executes a method of expressing semantic definitions on the computing device 100 searches for the existence of a hypernym / hyponym, whole word / part word, or concrete action word relationship including the intended semantic element or semantic concept, traverses the indicated number of downward levels accordingly, and extends the semantic concept relationship for each applicable semantic element or semantic concept.

[0032] In an exemplary embodiment, an N-level downward semantic inheritance statement for semantic concept relationships has the form "つ>>n [A] :: [X], [Y], [Z];", where "つ>>n" is a semantic inheritance operator, ">>" indicates the downward direction, and "n" indicates the number of levels to traverse. For example, if there is a statement "[Fruit] ::> [Grape], [Tomato], [Oyster];", the statement "つ>>1 [Fruit] :: [Antibacterial agent];" generates a one-level downward semantic inheritance such that each of the lower-level words of "Fruit", namely "Grape", "Tomato", and "Oyster", inherits the semantic concept relationship with the semantic element "Antibacterial agent".

[0033] Similarly, semantic context relationships can be extended N levels downward to the children of semantic elements or semantic concepts. This can be done by indicating the downward direction and the inheritance level with a semantic inheritance operator and the intended semantic element or semantic concept. The system searches for the existence of upper-level / lower-level words, whole words / partial words, or specific action word relationships that include the intended semantic element or semantic concept, traverses the indicated number of downward levels accordingly, and extends the semantic context relationship for each applicable semantic element or semantic concept.

[0034] In an exemplary embodiment, an N-level downward semantic inheritance statement for semantic context relationships has the form "つ>>n [A] ミ [B];", where "つ>>n" is a semantic inheritance operator, ">>" indicates the downward direction, and "n" indicates the number of levels to traverse. For example, if there are statements "[Vegetable] ::> [Cruciferous vegetable], [Root vegetable];" and "[Cruciferous vegetable] ::> [Broccoli], [Cauliflower];", the statement "つ>>2 [Vegetable] ミ [Anti-inflammatory];" generates a two-level downward semantic inheritance from "Vegetable" such that the first-level lower-level words of "Vegetable", namely "Cruciferous vegetable" and "Root vegetable", and the second-level lower-level words of "Vegetable" via "Cruciferous vegetable", namely "Broccoli" and "Cauliflower", inherit the semantic context relationship with the semantic concept "Anti-inflammatory".

[0035] In another exemplary embodiment, a semantic inheritance statement that is N levels downward with respect to a semantic context relationship has the form of "つ>>n [A] ミ [B] :: [X], [Y], [Z];". For example, if there are statements such as "[Vegetables] ::> [Cruciferous vegetables], [Root vegetables];" and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower];", then "つ>>2 [Vegetables] ミ [Health] :: [Anti - inflammatory];" generates a semantic inheritance that goes two levels down from "Vegetables" such that the first - level subordinate words of "Vegetables", namely "Cruciferous vegetables" and "Root vegetables", and the second - level subordinate words of "Vegetables" via "Cruciferous vegetables", namely "Broccoli" and "Cauliflower", inherit the semantic context relationship by the semantic concept "Health" described by the semantic element "Anti - inflammatory" as members of the semantic context relationship. Note that in this example, only the shown semantic element "Anti - inflammatory" is inherited. If there is a semantic context statement "[Vegetables] ミ [Health] :: [Antibacterial agent];", but it is not shown in the semantic inheritance statement, it is not inherited.

[0036] The semantic concept relationship can be extended downward for all available levels up to the children of the semantic element or semantic concept. This can be done by indicating the downward direction as well as all levels by the semantic inheritance operator and the intended semantic element or semantic concept. The system searches for the existence of superordinate / subordinate, whole / part, or specific action word relationships that include the intended semantic element, traverses all downward levels, and extends the semantic concept relationship for each applicable semantic element or semantic concept.

[0037] In an exemplary embodiment, a semantic inheritance statement that descends all levels for a semantic concept relationship has the form "つ>>a [A] :: [X], [Y], [Z];", where "つ>>a" is a semantic inheritance operator, ">>" indicates the downward direction, and "a" indicates crossing all levels. For example, if there are statements such as "[Plant-derived food] ::> [Vegetables], [Tofu], [Grains];", "[Vegetables] ::> [Cruciferous vegetables], [Root vegetables];", and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower];", then "つ>>a [Plant-derived food] :: [Anti-inflammatory];" generates semantic inheritance for all downward levels such that "Vegetables", "Tofu", "Grains", "Cruciferous vegetables", "Root vegetables", "Broccoli", and "Cauliflower" all inherit the semantic concept relationship with the semantic element "Anti-inflammatory".

[0038] Similarly, a semantic context relationship can be extended downward for all available levels down to the children of a semantic element or semantic concept. This can be done by indicating the downward direction as well as all levels by means of a semantic inheritance operator and the intended semantic element or semantic concept. The system searches for the presence of a hypernym / hyponym, whole / part, or specific action word relationship that includes the intended semantic element or semantic concept, crosses all downward levels, and extends the semantic concept relationship for each applicable semantic element or semantic concept.

[0039] In an exemplary embodiment, a semantic inheritance statement that descends all levels with respect to a semantic context relationship has the form "つ>>a [X] ミ [A], [B], [C];", where "つ>>a" is a semantic inheritance operator, ">>" indicates the downward direction, and "a" indicates crossing all levels. For example, if there are statements such as "[Plant-derived food] ::> [Vegetables], [Tofu], [Grains];", "[Vegetables] ::> [Cruciferous vegetables], [Root vegetables];", and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower];", then "つ>>a [Plant-derived food] ミ [Health] :: [Anti-inflammatory];" generates semantic inheritance for all downward levels of "Plant-derived food" such that "Vegetables", "Tofu", "Grains", "Cruciferous vegetables", "Root vegetables", "Broccoli", and "Cauliflower" all inherit the semantic context relationship by the semantic concept "Health" described by the semantic element "Anti-inflammatory" as members of the semantic context relationship.

[0040] Similarly, a semantic concept relationship can be extended N levels upward to the parent of a semantic element or semantic concept. This can be done by indicating the upward direction and the inheritance level with the semantic inheritance operator and the intended semantic element or semantic concept. The system searches for the existence of a hypernym / hyponym, whole / part, or specific action word relationship that includes the intended semantic element or semantic concept, crosses the indicated number of upward levels, and extends the semantic context relationship for each applicable semantic element or semantic concept. In some embodiments, the relationship may be further extended to members of the parent such as the parent's hyponyms, parts, or specific action words.

[0041] In an exemplary embodiment, an N-level upward semantic inheritance statement for semantic concept relationships has the form "つ<<n [A] :: [X], [Y], [Z];" where "つ<<n" is the semantic inheritance operator, "<<" indicates the upward direction, and "n" indicates the number of levels to traverse. For example, if there is a statement "[Berry] ::> [Grape], [Raspberry], [Blueberry];" then "つ<<1 [Grape] :: [Antibacterial agent];" generates a one-level upward semantic inheritance such that the superordinate word of "Grape", which is "Berry", inherits the semantic concept relationship with the semantic element "Antibacterial agent". In some embodiments, the upward inheritance includes siblings or other children of the parent such that "Raspberry" and "Blueberry" also inherit "Antibacterial agent".

[0042] Similarly, semantic context relationships can be extended N levels upward to the parent of a semantic element or semantic concept. This can be done by indicating the upward direction and the inheritance level with the semantic inheritance operator and the intended semantic element or semantic concept. The system searches for the existence of superordinate / subordinate word, whole / part word, or concrete action word relationships that include the intended semantic element or semantic concept, traverses the indicated number of upward levels, and extends the semantic context relationship for each applicable semantic element or semantic concept.

[0043] In an exemplary embodiment, the N-level upward semantic inheritance sentence for the semantic context relationship is "tsu< <n [A] ミ [B];」の形式を有し、ここで、「つ<<n」は意味継承演算子であり、「<<」は上向き方向を示し、「n」は横断するレベルの数を示す。例えば、「[野菜] ::> Given the sentences "[Cruciferous vegetables], [root vegetables];" and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower];", "つ<<2 [Cauliflower] ミ [anti-inflammatory];" generates a semantic inheritance two levels up from "cauliflower", such that "Cruciferous vegetables", the first-level hypernym of "cauliflower", and "vegetables", the second-level hypernym of "cauliflower" via "Cruciferous vegetables", inherit the semantic context relationship with the semantic concept "anti-inflammatory".

[0044] In another exemplary embodiment, a semantic inheritance sentence that goes up N levels for a semantic context relationship is<n [A] ミ [B] :: [X], [Y], [Z];」の形式を有する。例えば、「[野菜] ::> If the sentences "[Cruciferous vegetables], [Root vegetables];" and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower];" are present, then "つ<<2 [Cauliflower] ミ [Health] :: [Anti-inflammatory];" generates a semantic inheritance two levels up from "Cauliflower" such that "Cruciferous vegetables", the first-level hypernym of "Cauliflower", and "vegetables", the second-level hypernym of "Cauliflower" via "Cruciferous vegetables", inherit a semantic context relation with the semantic concept "health" along with the semantic element "anti-inflammatory" as members of the semantic context relation. Note that only the indicated semantic element "anti-inflammatory" is inherited. If the semantic context sentence "[Cauliflower] ミ [Health] :: [Antibacterial];" is present but not indicated in the semantic inheritance sentence, then "antibacterial" is not inherited.

[0045] The semantic concept relationship can be extended upward for all available levels up to the parent of the semantic element or semantic concept. This can be done by the semantic inheritance operator and the intended semantic element or semantic concept, indicating the upward direction as well as all levels. The system searches for the existence of hypernym / hyponym, meronym / holonym, or specific action word relationships that include the intended semantic element or semantic concept, traverses all upward levels, and extends the semantic concept relationship for each applicable semantic element or semantic concept.

[0046] In an exemplary embodiment, the upward semantic inheritance statement for all levels of the semantic concept relationship has the form "つ<<a [A] :: [X], [Y], [Z];" where "つ<<a" is the semantic inheritance operator, "<<" indicates the upward direction, and "a" indicates traversing all levels. For example, if there are statements such as "[Plant-derived food] ::> [Vegetables], [Tofu], [Grains];", "[Vegetables] ::> [Cruciferous vegetables], [Root vegetables];", and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower], [Cabbage];", then the statement "つ<<a [Cabbage] :: [Anti-inflammatory];" generates semantic inheritance for all upward levels so that "Cruciferous vegetables", "Vegetables", and "Plant-derived food" all inherit the semantic concept relationship with the semantic element "Anti-inflammatory".

[0047] Similarly, the semantic context relationship can be extended upward for all levels up to the parent of the semantic element or semantic concept. This can be done by the semantic inheritance operator and the intended semantic element or semantic concept, indicating the upward direction as well as all levels. The system searches for the existence of hypernym / hyponym, meronym / holonym, or specific action word relationships that include the intended semantic element or semantic concept, traverses all upward levels, and extends the semantic context relationship for each applicable semantic element or semantic concept.

[0048] In an exemplary embodiment, all levels of upward semantic inheritance statements have the form "つ<<a [X] ミ [A], [B], [C];", where "つ<<a" is the semantic inheritance operator, "<<" indicates the upward direction, and "a" indicates crossing all levels. For example, if there are statements such as "[Plant-derived food] ::> [Vegetables], [Tofu], [Grains];", "[Vegetables] ::> [Cruciferous vegetables], [Root vegetables];", and "[Cruciferous vegetables] ::> [Broccoli], [Cauliflower], [Cabbage];", then "つ<<a [Cabbage] ミ [Health] :: [Anti-inflammatory];" generates the upward semantic inheritance of all levels of "Cabbage" such that each of the semantic concepts "Cruciferous vegetables", "Vegetables", and "Plant-derived food" inherits the semantic context relationship with the semantic element "Anti-inflammatory" by the semantic concept "Health" as members of the semantic context relationship. In one embodiment, the semantic context relationship is further extended to the subordinate word members of "Cruciferous vegetables" ("Broccoli" and "Cauliflower"), "Vegetables" ("Cruciferous vegetables" and "Root vegetables"), and "Plant-derived food" ("Vegetables", "Tofu", and "Grains").

[0049] Semantic instances further define semantic elements or semantic concepts with specific content. For example, the semantic element "[Winery]" can have a semantic instance that is the exact name of a winery such as "[Deer Hills Winery]". Semantic instances may be defined by a user having authoring rights in a semantic definition management editor. In the semantic definition management editor, semantic instances can be selected from a menu as single selection, multiple exclusive selection (OR condition), and / or multiple non-exclusive selection (AND condition).

[0050] In an exemplary embodiment, a semantic instance sentence has a general form of "[A] yu {B};", where A is a semantic element or semantic concept, "yu" is a semantic instance operator, and B is a semantic instance such as "[Winery] yu {Diablo Hills Winery}". A semantic instance sentence can have a form of "[A] yu {B}, {C}" for multiple non-exclusive selections (AND conditions) such as "[Taste] yu {Sweetness}, {Sourness}, {Bitterness};", where a user can select any combination of "Sweetness", "Sourness", and "Bitterness" as semantic instances of "Taste". A semantic instance sentence having multiple exclusive selections (OR conditions) can have a form of "[A] yu {B} | {C};" such as "[Fiber] yu {High} | {Low};", where a user can select only one of "High" or "Low" as a semantic instance of "Fiber". To ensure that a semantic instance is not defined by a user during visual definition processing, a semantic instance sentence may have a form of "[A] yu 0;", for example, "[Grape] yu 0;" prevents a user from defining a semantic instance of "Grape". In a graphical user interface, multiple non-exclusive selections may be presented to the user as check boxes, and multiple exclusive selections may be presented to the user as radio buttons during visual processing and / or visual search.

[0051] A semantic instance may be represented in another language or notation system via a semantic instance internationalization definition represented by a semantic instance internationalization sentence. A semantic instance internationalization sentence includes a semantic instance in a first language or notation system, a semantic instance internationalization operator, and a translation of the semantic instance into a second language or notation system. A user can select a language via a language switch of the graphical user interface to determine the language to be used during search or other processing.

[0052] In an exemplary embodiment, a semantic instance internationalization statement has the general form "@aa{semantic instance X} u @xx{translated semantic instance X}", where "u" is the semantic instance internationalization operator, "@aa" is a language code operator that includes a two-letter language code that identifies the original language of the semantic instance, and "@xx" is a language code operator that includes a two-letter language code that identifies the translation language of the semantic instance. For example, "@en{Grape} u @de{Weintraube}, @es {Uva};" represents a semantic instance internationalization relationship in which the English ("en") semantic instance "Grape" is translated into the German ("de") semantic instance "Weintraube", and the Spanish ("es") semantic instance "Uva".

[0053] For languages with multiple writing systems, such as Japanese, semantic instance internationalization can be defined per writing system. The default writing system for a language can be selected from among the multiple writing systems.

[0054] In an exemplary embodiment, a semantic instance internationalization statement indicating different writing systems for a single language uses a writing system operator of the form "@xx-yy", where "xx" identifies the language and "yy" identifies the writing system. A semantic instance internationalization statement may use "!" as the default writing system operator. For example, the semantic instance internationalization statement "@en{Grape} !@jp-hg{budou}, @jp-kj{budou}, @jp-kk{budou}, @jp-rj{budou};" indicates that the English semantic instance "Grape" can be translated into Japanese in four different writing systems: Hiragana, abbreviated as "hg", Kanji, abbreviated as "kj", Katakana, abbreviated as "kk", and Romaji, abbreviated as "rj". Hiragana has been selected as the default writing system by the default writing system operator, so that when the user interface language preference is Japanese, results are returned to the user in Hiragana without the need to specify a writing system.

[0055] The semantic internationalization definition may be used to translate semantic elements and concepts into other languages and writing systems. Semantic internationalization sentences may be formed substantially similarly to semantic instance internationalization sentences, as described above. In an exemplary embodiment, a semantic internationalization sentence has the general form "@aa[semantic element / semantic concept] @xx[translated semantic element / semantic concept]", where "@" is the semantic internationalization operator (which may be the same as the semantic instance internationalization operator), and "@aa" and "@xx" are the respective language code operators for the original language and the translated language (which may be the same as the language code operators for the semantic instance internationalization sentence). For example, "@en{Grape} @de{Weintraube}, @es {Uva};" represents a semantic internationalization relationship in which the English semantic concept "Grape" is translated into the German semantic concept "Weintraube" and the Spanish semantic concept "Uva".

[0056] Semantic elements or concepts can be translated into different writing systems in a manner substantially similar to semantic instances, as described above. In an exemplary embodiment, a semantic internationalized sentence for translation into different writing systems has the general form "@aa[semantic instance X] @xx-yy[translated semantic instance written in writing system yy of language xx]". For example, "@en[Grape] @jp-hg[budou], @jp-kj[budou], @jp-kk[budou], @jp-rj[budou];" represents that the English semantic element "Grape" can be translated into Japanese in four different writing systems, with Hiragana being selected as the default writing system using the default writing system operator "!".

[0057] The lexical function defines synonyms, antonyms, and grammatical forms of semantic elements and semantic concepts. The lexical function definition can be represented as a lexical function sentence that includes a semantic element or semantic concept, a lexical function operator, and synonyms, antonyms, or grammatical forms of the semantic element or semantic concept, depending on the lexical function operator.

[0058] In an exemplary embodiment, a synonym lexical function sentence has the general form "[A] α [synonym];", where "α" is the synonym lexical function operator. For example, "[Vegetables] α [Veggies];" represents that "Vegetables" and "Veggies" are synonyms. The system can treat synonyms interchangeably so that any semantic relationship involving "Vegetables" is also applicable to "Veggies".

[0059] In an exemplary embodiment, an antonym lexical function sentence has the general form "[A] >< [B]", where "><" is the antonym lexical function operator. For example, "[North] >< [South]" represents that "North" and "South" are antonyms.

[0060] The grammatical form of a semantic element or semantic concept may include singular / plural forms. In an exemplary embodiment, a singular / plural lexical function sentence has the general form "[A] =sp= [B];", where "=sp=" is the singular / plural lexical function operator indicating that "A" is in the singular form and "B" is in the plural form. For example, "[Grape] =sp= [Grapes];" represents that "Grape" has the plural form "Grapes".

[0061] The grammatical form of a semantic element or semantic concept may include a word form transformation. A word form transformation is similar to a synonym, but the difference is that the word form transformation can be further defined by grammar rules. In an exemplary embodiment, the vocabulary function statement of a word form transformation has a general form of "[A] == [B];", where "==" is a vocabulary function operator of a word form transformation indicating that "A" and "B" are different word forms of the same word. For example, "[variable] == [varietal];" represents that "variable" and "varietal" are different word forms of the same semantic element or semantic concept. A word form transformation may be restricted to a specific semantic domain. This can be represented in a general form of "[A] == [B] in <<semantic domain>>;". For example, "[variable] == [varietal] in <<botany>>;" represents that "variable" and "varietal" are word form transformations of each other only in the semantic domain of "botany".

[0062] For each semantic definition, the author can, and preferably does, provide a reference to a reliable source, such as a URL (Uniform Resource Locator) that references the source of the knowledge / information supporting the semantic definition. The reference may include a brief description of the source. The source can be the author's personal website or another external website. One reference or multiple references can be applied to one semantic definition statement, or the reference can be applied to a complete metaset.

[0063] A semantic reference definition may be added to any semantic definition statement using the semantic reference operator. In an exemplary embodiment, the semantic reference statement has the form “[A] :: [B]; ma URL ((description));”, where “[A] :: [B];” is any of the above-described types of semantic definition statements, “ma” is the semantic reference operator, “URL” is the URL linking to the information source, and “((description))” is the free text description of the information source. For example, “[Cabbage] :: [Anti-inflammatory]; ma http: / / www.xyz.org ((XYZ Health Institute, July 2010))” indicates that the semantic concept relationship between “cabbage” and “anti-inflammatory” is supported by the information at http: / / www.xyz.org available from the XYZ Health Institute as of July 2010. In another example, a second source is provided as “[Cabbage] :: [Anti-inflammatory]; ma http: / / www.xyz.org ((XYZ Health Institute, July 2010)), http: / / www.cde.com ((CDE Association, August 2016));”.

[0064] The semantic reference definition applied to the complete metaset includes a reference to the metaset, the semantic reference operator, and a citation to the source. In an exemplary embodiment, the statement containing the semantic reference definition applied to the complete metaset has the general form “(s) ma URL ((description));”, where “(s)” refers to the metaset being currently edited. For example, “(s) ma http: / / www.xyz.org ((XYZ Health Institute, July 2010)), http: / / www.cde.com ((CDE Association, August 2016));” represents two sources applied to the metaset being currently edited. In other embodiments, the metaset may be referred to by a file name, a URL, or other reference indicator.

[0065] Referring to FIG. 1, in step 120, the meaning definition text is stored within a collection known as a metaset. In one embodiment, the metaset is a database. In other embodiments, the metaset may be any type of file capable of storing text and operators. The metaset may have a unified semantic theme such as "fruit" or "fish". The metaset can be edited by a specific author or by any number of authors with editing rights. An example of a metaset may be any collection of meaning definition texts related to the above fruits and vegetables.

[0066] In step 130, the metaset is converted into a digital data structure. In step 140, the digital data structure is stored in the memory storage device of the computer device. In some embodiments, the memory storage device may be a hard disk. In other embodiments, the memory storage device may be RAM, cloud storage, a USB drive, a CD, or any other memory storage device known in the art. Various embodiments of a computer system capable of executing a method for representing meaning definitions on the computing device 100 will be described below with reference to FIG. 7.

[0067] FIG. 2 shows a meaning definition management editor 200 according to an embodiment of the present invention. The meaning definition management editor may be used to compose the meaning definition text 210 of the metaset 220 as in step 110.

[0068] The editor can be improved by a set of preconfigured meaning definition text templates 230, providing readily usable meaning definition texts for users with authoring rights to define and assign semantic relationships as a metaset. The templates may include template meaning definition texts that assist in defining semantic concepts, semantic contexts, semantic labels, semantic domains, semantic inheritance, semantic instances, vocabulary functions, semantic references, and semantic internationalization.

[0069] As part of creating a metaset in the meaning definition management editor, the user may be required to describe the semantics of the metaset. The semantics of a metaset may include a title such as "Fish and Health", semantics in the form of key:value pairs such as "Animal: Fish", and a free text description of the metaset such as "Fish and its health benefits". The key:value pairs enable convenient organization of the metaset, and the free text description may enable the user to easily include information that the user considers relevant to the metaset.

[0070] The meaning definition management editor 200 has functions such as open, create, save, save as, etc. for managing metasets. When a metaset is saved, it is converted into a data structure and saved in a digital format that can be used for editing, querying, and searching during meaning definition and visual semantic search operations.

[0071] When a metaset is saved, automatic recognition of a meaning definition text including meaning concepts (general, superordinate / subordinate words, whole / part words, concrete action words, free), semantic contexts (general, free), semantic labels (general, semantic label collection), semantic domains (general, label-domain, domain-domain, domain-transcendent domain), semantic inheritance, semantic instances, vocabulary functions, semantic references, and semantic internationalization sentences is triggered. Automatic recognition of the meaning definition text can identify the type of the meaning definition text and the components of the meaning definition text. The meaning definition management editor may provide feedback indicating that the recognition of the meaning definition text was successful or an error in the meaning definition text.

[0072] Editing of the metaset via the semantic definition management editor 200 is available during visual processing or outside visual processing. During visual processing, the creator of the metaset used in visual semantic processing (or, in the case of a non-creator, for a metaset marked as publicly editable or selectively editable by the current user having the right to change the metaset) can pause or temporarily end the visual semantic processing in order to edit the metaset. When the editing is completed (by the user's save operation) in the semantic definition management interface, the computer system executing the semantic definition management editor 200 verifies the changes made to the metaset. When the verification is completed, the user is redirected to the visual semantic processing interface from which the user has left. Thereafter, the user can continue the visual semantic processing using the changed metaset.

[0073] Outside visual processing, the user accesses the semantic definition management editor 200, searches for the metaset file name, selects the file name from the list of metasets, and opens the metaset using the open function within the interface. When the metaset is opened, the metaset is displayed in the online editor and the creator can edit and save the metaset. Then the system verifies the changes and checks that the format is fully compliant before updating the data structure of the metaset.

[0074] All saved metasets are compiled and published as a playlist within a metaset library with attributes / descriptors such as title, metaset name, creator, user name, semantics, description, and last update date, and are available for use and search by the user (in the case of a user having such rights) to search for, detect, and edit the metaset. The metaset library can be displayed within the window of the semantic definition management editor or separately. Furthermore, the metaset library playlist can provide a function to directly display the content of the metaset by a click or one operation on the user interface without the need to access the online editor.

[0075] Figure 3 shows a method for semantically searching visual 300 according to an embodiment of the present invention. When a visual is processed (for example, as described in Japanese Patent No. 6830514), the visual can be semantically searched.

[0076] In step 310, the input is received via a computer interface. The input may include text. For example, text terms may include "fish" and "sardine". The input may include a selection of multiple options from a graphical user interface. For example, a selection can be made to show the semantic relationship between "fish" and "sardine". The input may include an image such as a picture of a sardine to find detailed information or semantics of an input image or a visual similar to the input image. The input may be received from a local system or a remote system via a network.

[0077] In step 320, a semantic representation is derived from the input. The semantic representation conveys semantic meaning in a form that may be different from the form of a semantic definition sentence. The semantic representation may be the same as the text input. The semantic representation may be a simple keyword. The semantic representation may have a strict form of semantic definition sentence as described above. The semantic representation may be text derived from an input visual via visual processing. The semantic representation can represent all or part of the semantic definition. The semantic representation may be a framework of semantic definition sentences that can be intuitively used by the user. For example, the semantic representation can associate keywords or phrases by a selection of multiple options from a user interface, and the keywords or phrases may correspond to operands of a semantic definition sentence. The user can also select an item from a menu corresponding to an operator of the semantic definition sentence. In the above example, the text inputs "fish" and "sardine" are used, together with a selection of multiple options indicating that a sardine is a type of fish, to derive a semantic representation showing the superordinate / hyponym semantic concept relationship between "fish" and "sardine".

[0078] In step 330, at least one type of semantic definition is determined to be applied to visual search. The types of semantic definitions can include any of the above-mentioned semantic definitions such as semantic element definition, semantic concept definition, semantic context definition, semantic label definition, semantic area definition, semantic inheritance definition, semantic instance definition, semantic internationalization definition, vocabulary function definition, and semantic reference definition. For example, the search may be performed only according to the semantic element definition, semantic concept definition, and semantic context definition. The determination of which type of semantic definition to apply may be based on user input. This determination may alternatively be automatically applied by a computer system that executes a method of semantically searching the visual 300.

[0079] In step 340, the search is performed based on the input, such as by using a semantic representation and at least one type of semantic definition determined to be applied to the search. Further, the search may be based on an "active" metaset selected by the user. The metaset created by the above method can be used to apply semantic definitions to the search, and the type of semantic definition is selected by the user and applied to the input or semantic representation derived from the input. The active metaset can be selected from a user interface where the metaset may be presented by attributes such as metaset name, title, semantics, description, creator, authorized users, or last update date for searching, sorting, or other means. For example, a "fish" metaset containing semantic definitions related to visuals of various types of fish can be applied to the search for "sardine" as a hyponym of "fish".

[0080] The search can be performed such that the semantic definition is required to match all semantic expressions like both "fish" and "sardine", or any subset of semantic expressions like either "fish" or "sardine". A strict semantic search can be selected such that the semantic expression is required to exactly match the form of the semantic definition in order to be considered a match. Alternatively, under a loose semantic search, the semantic expression may match the terms of the semantic definition regardless of the exact form of the semantic definition that is considered a match. For example, when "fish" and "sardine" are input as having a general semantic concept relationship, a visual having a semantic definition that "sardine" is a hyponym of "fish" matches the input under a loose semantic search but does not match under a strict semantic search. Alternatively, a pivot search may tightly match a specific semantic expression to the semantic definition while loosely matching with other semantic expressions. For example, a pivot search may require that "sardine" be strictly searched as a hyponym of "fish", while on the other hand, "tuna" may match a semantic definition that does not specify "tuna" as a hyponym of "fish". The search can be specified to include all visuals available for the search, or only "My Visuals", which are the visuals processed by the logged-in user performing the search.

[0081] In step 350, a list of visuals determined to have semantic relevance to the input based on the search is output. In the above example, the list of visuals may include an image including sardines. The user may select a visual from the list of visuals to view the visual or to view the semantic definition associated with the visual.

[0082] FIG. 4 shows a method of selecting a semantic definition from a metaset 400 according to an embodiment of the present invention. To enable a user to select a semantic definition from a metaset, there are several applications such as navigating through the metaset (FIG. 5) and filtering the metaset (FIG. 6).

[0083] In step 410, the semantic definition sentence is retrieved from one or more meta-sets. As described above, the semantic definition sentence may represent a semantic definition and have a general form of a subject operand, an operator, and at least one object operand.

[0084] In step 420, a set of semantic expressions based on the semantic definition sentence is output. The set of semantic expressions includes selectable items such as selectable operands and operators, or items that are intuitively usable by the user corresponding to operands and operators such as images and emojis. The output may be displayed on a graphical user interface.

[0085] In step 430, a selection of a selectable item is received. The selection may be made by the user in a graphical user interface.

[0086] In step 440, the selected semantic definition is determined based on the set of semantic expressions and the selection of the selectable item. This can be done by matching the corresponding operator and operand of the semantic definition sentence with the semantic expression and the selection of the selectable item.

[0087] FIG. 5 shows a user interface 500 for navigating a metaset according to an embodiment of the present invention. As can be seen in the user interface 500, the collapsible menu 510 is output such that the semantic expressions 520 corresponding to the operands of the meaning definition sentences from the metaset are selectable (step 420). The user selects a desired selection 530 from the semantic expressions 520 for navigation. The computer system executing the user interface 500 receives the selection 530 (step 430), determines the corresponding meaning definition (step 440), and accordingly updates the collapsible menu 510, such as by expanding the selection 530 with additional menu items 540 below. The collapsible menu 510 is then updated by collapsing the additional menu items 540 below the selection 530. The visual 550 may be provided to display one or more visuals related to the selection 530. For example, if the selection 530 corresponds to the semantic concept "fish", the submenu of the additional item 540 is expanded such that different types of semantic concepts such as "overview" and "hyponyms" are selectable. Further selecting "hyponyms" expands another submenu listing the hyponyms of "fish" such as "sardine". In this way, the metaset can be navigated.

[0088] Figure 6 shows a user interface 600 for filtering a metaset according to an embodiment of the present invention. As can be seen in the user interface 600, the menu 610 outputs a list of semantic expressions 620 corresponding to the subject operand, and selectable semantic expression pairs 630 corresponding to the object operand 632 and the operator 634 (step 420). For each semantic expression pair 630 that the user desires to use to filter the metaset, the user can select the check box 640. The visual 650 may be one or more visuals included to guide the user's selection. Additionally, or alternatively, text or audio may be included to guide the user's selection. The computer system that executes the user interface 600 receives the selection (step 430) and filters the metaset for excluding the semantic definition corresponding to the semantic expression 620 and the semantic expression pair 630 with the unselected check box 640 (step 440). For example, the semantic expression 620 may correspond to the semantic concept "fish". The semantic expression pair 630 can include a fish species as the object operand 632, and can indicate that the fish species is a hyponym as the operator 634. The user can recognize the sardine by looking at the visual 650. Then the user selects only the check box 640 corresponding to "sardine". The system receives this selection that only "sardine" is a hyponym of "fish", and filters the metaset to exclude "ohyo", "tuna", and "salmon" as hyponyms of "fish".

[0089] Figure 7 shows a computer system 700 according to an embodiment of the present invention. The computer system 700 includes a client computer 710 and a server computer 720 connected by a network 730. The network 730 can be any type of network such as a local area network or the Internet. In an alternative embodiment, only a single client computing device may be required.

[0090] The client computer 710 includes a memory 711, a CPU 712, an input device 713, an output device 714, and a network interface 715. The client computer 710 can be a personal computer, a tablet PC, a smartphone, a laptop computer, or other well-known computing devices. The memory 711 may be a non-transitory medium for storing instructions for executing any of the above-described methods. The memory 711 may include RAM, a hard disk, a flash drive, or any other memory storage device well-known in the art. In some embodiments, the memory 711 is used to store image files, meta-sets, digital data structures, program software for visual processing and visual search, or any other data desirable for executing the above-described methods. The memory 711 may include removable or portable memory devices such as CDs, USB drives, external hard drives, etc. The instructions stored in the memory 711 can be executed by the CPU 712. The instructions may be additionally or alternatively processed by another processor of the client computer 710 or by another processor of a client-side device communicating with the client computer 710. The user can provide inputs via an input device 713 such as a keyboard, a mouse, a scanner, a webcam, etc. Outputs such as the above-described meaning definition editor or other graphical user interfaces may be displayed on the output device 714, which may be a monitor. The network interface 715 enables communication with the server computer 720 via the network 730. For example, a user of the client computer 710 can send a visual search request to the server computer 720 and receive the results of the search request from the server computer 720 via the network 730.

[0091] The server computer 720 includes a memory 721, a CPU 722, and a network interface 723. The memory 721 may include one or more databases for storing visualizations and meta-sets that describe the visualizations. The memory 721 can store a visual processing or visual search program as instructions executable by the CPU 722. The memory 721 may have a considerably larger storage capacity than the memory 711 of the client computer 710. Similarly, the CPU 722 may have a considerably larger processing capacity than the CPU 712 of the client computer 710. The CPU 722 may be configured to execute instructions for visual processing, visual search, or other requests in parallel from various client computers according to the methods described above. The server computer 720 can send the results of any processing to the client computer 710 or any other appropriate target via the network 730.

[0092] Although the present disclosure has been specifically shown and described with reference to preferred embodiments and various aspects thereof, it will be understood by those skilled in the art that various changes and modifications can be made without departing from the spirit and scope of the present disclosure. The appended claims are intended to be construed to include the embodiments described herein, the alternatives described above, and all equivalents thereof.

Explanation of Signs

[0093] 100 Method for Representing Meaning Definitions on a Computing Device 110 Composing a Meaning Definition Sentence 120 Saving a Meaning Definition Sentence in a Meta-Set 130 Converting a Meta-Set into a Digital Data Structure 140 Storing a Digital Data Structure in a Memory Storage Device of a Computing Device 200 Meaning Definition Management Editor 210 Meaning Definition Sentence 220 Meta-Set Set of pre - configured templates Method for visually semantically searching Receiving an input via a computer interface Deriving a semantic representation from the input Determining at least one type of semantic definition to apply to the search Executing a search based on the input and at least one type of semantic definition Outputting a list of determined visuals having semantic relevance to the input based on the search Method for selecting a semantic definition from a metaset Extracting a semantic definition sentence representing a semantic definition from one or more metasets Outputting a set of semantic representations based on the semantic definition sentence and including selectable items Receiving a selection of a selectable item Determining the selected semantic definition based on the set of semantic representations and the selection User interface for navigating a metaset Folding menu Semantic representation Selection Additional menu items Visual User interface for filtering a metaset Menu Semantic representation Selectable semantic representation pairs Object operand Operator Checkbox Visual Computer system Client computer Memory CPU 713 Input Device 714 Output Device 715 Network Interface 720 Server Computer 721 Memory 722 CPU 723 Network Interface 730 Network

Claims

1. A method for representing a semantic definition on a computer device, the computer device comprising: receiving an input from a user; in accordance with the received input, editing and displaying a semantic definition sentence that includes a semantic concept, a semantic concept operator, and at least one semantic element and / or free text description having a semantic concept relationship with the semantic concept, and defining the semantic concept, and a first semantic context operator and at least two semantic concepts having a semantic context relationship with each other, and including a semantic context sentence defining a semantic relationship between the at least two semantic concepts; storing the semantic definition sentence as a metaset that is a dataset stored in a text and operator-arranged sentence format; storing the metaset in a memory storage device and executing the method.

2. The method according to claim 1, wherein the semantic concept operator is a general semantic concept operator, a superordinate word operator, a subordinate word operator, a whole word operator, a part word operator, a specific action word operator, or a free text operator.

3. The method according to any one of claims 1 or 2, wherein at least one of the semantic context sentences further includes a second semantic context operator for associating at least one semantic element and / or free text description with a semantic context relationship.

4. The method according to any one of claims 1 to 3, wherein the semantic definition sentence further includes a semantic label sentence including a semantic label operator and a semantic label that respectively describe one of the semantic concept relationships or one of the semantic context relationships in more detail than the semantic concept relationship or the semantic context relationship.

5. The semantic definition sentence includes one of the semantic concepts, a semantic label collection operator, and a group of the semantic labels and further includes a semantic label collection sentence respectively including them, and each semantic label of the group of semantic labels is found in at least one of the semantic label sentences including one of the semantic concepts. The method according to claim 4.

6. The method according to any one of claims 1 to 5, wherein the meaning definition sentence further includes a meaning area sentence including a meaning area operator, a meaning area, at least one meaning element, meaning concept, semantic label, or another meaning area assigned to the meaning area.

7. The method according to any one of claims 1 to 6, wherein the meaning definition sentence includes a meaning inheritance operator and a meaning concept sentence or a semantic context sentence, and further includes a meaning inheritance sentence for inheriting at least one meaning concept relationship or semantic context relationship indicated by the meaning concept sentence or the semantic context sentence to at least one meaning element or meaning concept.

8. The method according to claim 7, wherein the meaning inheritance operator includes an inheritance level indicator indicating the number of levels of meaning elements or meaning concepts inheriting the meaning concept relationship or semantic context relationship, and / or an inheritance direction indicator indicating the direction of the level of meaning elements or meaning concepts inheriting the meaning concept relationship or semantic context relationship.

9. The method according to any one of claims 1 to 8, wherein the meaning definition sentence further includes a semantic instance sentence respectively including a meaning element or meaning concept, a semantic instance operator, and at least one semantic instance of the meaning element or the meaning concept.

10. The method according to claim 9, wherein the at least one semantic instance is selected from a single value, a plurality of non - exclusive values, a plurality of exclusive values, or a value indicating that the at least one semantic instance is undefined.

11. The method according to claim 9 or 10, wherein the semantic instance operator is a semantic instance internationalization operator for defining semantic instances of meaning elements or meaning concepts in different languages and / or different notation systems.

12. The method according to any one of claims 1 to 11, wherein the meaning definition sentence further includes a vocabulary function sentence including a meaning element or meaning concept, a vocabulary function operator, and at least one of a synonym, antonym, singular form, plural form, or variant of the same word of the meaning element or the meaning concept.

13. The method according to any one of claims 1 to 12, wherein the meaning definition sentence includes a meaning reference operator providing at least one reference for the meaning definition sentence.

14. The method according to any one of claims 1 to 13, wherein the semantic definition text further includes a semantic internationalization text including a semantic element or semantic concept, a semantic internationalization operator, and a conversion of the semantic element or semantic concept into different languages and / or writing systems.

15. The meta-set is associated with attributes including the title of the meta-set, the description of the meta-set, and / or the semantics of the meta-set configured as key:value pairs.

16. The step of displaying the semantic definition text in an editable manner and saving the semantic definition text as the meta-set is performed using a semantic definition management editor, and the semantic definition management editor is further configured to create a new meta-set, open a saved meta-set, edit the meta-set, save the meta-set with a new name, query the meta-set, and / or search for the meta-set.

17. The method according to claim 16, wherein the computer device further performs providing a set of pre-configured templates to the semantic definition management editor to assist in the composition of the semantic definition text.

18. The method according to claim 16 or 17, wherein the computer device further performs displaying the meta-set in an editable manner while using the meta-set for visual processing of obtaining semantic attributes from an image.

19. The method according to any one of claims 16 to 18, wherein the step of saving the semantic definition text as the meta-set triggers an automatic recognition by the semantic definition management editor of the type and components of the semantic definition text.

20. The semantic definition management editor is configured to create a list of meta-sets, and the meta-sets in the list of meta-sets are selectable for use during the composition of the semantic definition text, during visual processing of deriving text from an image, and / or during visual search for performing an image-based search.

21. The method according to any one of claims 16 to 19.

21. The list of the metasets is published as a metaset library accessible outside the semantic definition management editor, and the metasets in the metaset library are searchable by attributes associated with the metasets, according to the method of claim 20.

22. A method for semantically searching for visuals, comprising: a computer receiving an input including text or an image via a computer interface; deriving a semantic expression having a semantic meaning from the input; determining, based on user input, a type of semantic definition to apply to the search, wherein the type of semantic definition includes a semantic element definition that defines semantic elements including visual or non-visual attributes; a semantic concept definition that defines a semantic concept relationship between a semantic concept and a semantic element or another semantic concept; a semantic context definition that defines a semantic context relationship between semantic concepts; a semantic annotation definition that defines a semantic concept relationship or a semantic context relationship in more detail than the semantic concept relationship or the semantic context relationship; a semantic domain definition that assigns semantic elements, semantic concepts, semantic annotations, or other semantic domains to a semantic domain; a semantic inheritance definition in which a semantic element or a semantic concept inherits a semantic concept relationship or a semantic context relationship from another semantic element or semantic concept; a semantic instance definition that defines a semantic instance of a semantic element or a semantic concept; a semantic internationalization definition that defines a conversion of a semantic element or a semantic concept into a different language and / or a writing system; a vocabulary function definition that defines synonyms, antonyms, singular and plural forms, and / or inflections of the same word of a semantic element or a semantic concept, and a semantic reference definition that quotes a reference for one or more semantic definitions, determining the type of semantic definition; receiving a user's selection regarding a metaset that is a dataset storing a semantic definition sentence representing a semantic definition in a sentence format in which text and operators are arranged; performing a search for a semantic definition that matches the semantic expression based on the derived semantic expression, the determined type of semantic definition, and the selected metaset, and determining that at least one visual including an image and pre-associated with the semantic definition has a semantic relevance to the input. Outputting a list of visuals determined to have a semantic relationship with the input based on the search; A method of performing.

23. The step of performing the search is The method according to claim 22, wherein the search is performed by applying a semantic definition included in the selected metaset to the semantic expression.

24. The method according to claim 22 or 23, wherein the input is received from a logged-in user who performs the search, and the list of visuals is based on a search for a group of visuals processed by the logged-in user.

25. Determining that a visual has a semantic relationship with the input includes Determining that each of the semantic expressions matches a semantic definition pre-associated with the visual, Determining that each of a subset of the semantic expressions matches a semantic definition pre-associated with the visual, or Determining that any of the semantic expressions matches a semantic definition pre-associated with the visual The method according to any one of claims 22 to 24.

26. Determining that a visual has a semantic relationship with the input includes When a strict semantic search is selected, determining that the visual has a semantic relationship with the input only when it is determined that the semantic expression matches the form and terms of the semantic definition related to the visual. The method according to any one of claims 22 to 25.

27. The method according to any one of claims 22 to 26, further including outputting a semantic definition associated with the at least one visual when at least one visual is selected from the list of visuals.

28. A method of selecting a semantic definition from a metaset, wherein a computer Retrieving a semantic definition sentence representing a semantic definition from one or more metasets storing the semantic definition sentence in a text and sentence format with operators arranged, each semantic definition sentence including a subject operand, an operator, and an object operand; Outputting a set of semantic expressions based on the semantic definition text, where the set of semantic expressions includes the subject operand or the object operand and an item that can select the subject operand or the object operand, and outputting a set of semantic expressions; Receiving a selection of the selectable item; Determining a semantic definition corresponding to the set of semantic expressions for which the selectable item is selected A method for execution.

29. The method according to claim 28, further comprising filtering the one or more meta-sets based on the determined semantic definition.

30. The method according to claim 28 or 29, further comprising outputting an additional set of semantic expressions corresponding to the determined semantic definition.

31. A non-transitory computer-readable medium storing instructions executable by a processor for performing the method according to any one of claims 1 to 30.

Images