CN106980901A - Streaming RDF data parallel reasoning algorithm - Google Patents
Streaming RDF data parallel reasoning algorithm Download PDFInfo
- Publication number
- CN106980901A CN106980901A CN201710246309.2A CN201710246309A CN106980901A CN 106980901 A CN106980901 A CN 106980901A CN 201710246309 A CN201710246309 A CN 201710246309A CN 106980901 A CN106980901 A CN 106980901A
- Authority
- CN
- China
- Prior art keywords
- data
- rule
- triple
- node
- reasoning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/045—Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides streaming RDF data parallel reasoning algorithm：The pseudo- bilateral network of rule is built, intermediate node is set up if the link variable that there is class in regular node；The data that batch new data in timing acquisition Streaming data flows and previous reasoning are produced are as input data, and the data to input are sorted out or newly-built corresponding node and stored to corresponding Redis clusters；Judge whether the former piece that corresponding intermediate node or regular node are monitored all meets for the pseudo- bilateral network of triple data combination of input, and then the rule is made inferences, produce inference data；All data produced by deleting repetition inference data and this reasoning in real time are saved in the input data as next reasoning in Redis clusters, thus efficient realize the regular parallel streaming reasonings of RDF data OWL Horst.
Description
Technical field
The invention belongs to semantic network technology field, specifically related to streaming RDF data parallel reasoning algorithm.
Background technology
In recent years, researchers gradually recognize the importance of the parallel reasoning algorithm research of real-time streaming data, but are directed to
The related algorithm that the field is proposed is still less, needs further research.The also phase of research of the reasoning in terms of intellectual technology simultaneously
When many, such as discovery of knowledge, the reasoning of case etc..Extensive RDF stream datas phase is solved by Distributed Parallel Computing
Pass problem turns into the common recognition of academia and industrial quarters.
Research RDFS/OWL streaming parallel inferences are a newer at present fields.Barbieri D F et al. propose base
In the increment reasoning algorithm of streaming and rich background knowledge, the algorithm adds the temporal information that expires into each RDF triples, when new
Stream data when reaching, the fact that make inferences calculatings to new data, and terminate clear and definite and the invalid triple of deletion.
IDRM algorithms efficiently expansible can carry out RDFS reasonings to incremental data, because IDRM algorithms are special to RDFS rule progress
Modeling, so for the inefficient of OWL Horst rule-based reasonings.Chevalier J et al. put forward a kind of effective increment
Reasoning device（Slider）, the reasoning device is made inferences by the internal characteristicses in semantic data stream to it, it is achieved thereby that being directed to
The expansible batch processing reasoning device of stream data.But because Slider is designed just for RDFS rules, so for complexity
OWL Horst rule-based reasonings and do not apply to.
Nowadays being challenged present in extensive RDF files reasoning has：Distributed data on network is difficult to be fitted
When triple；Growing data volume requires the expansible computing capability of large data sets；Existing inference method is special
For static body, data are typically the change in real world.Existing distributing inference method primarily focuses on static number
According to reasoning, research streaming RDF data parallel inference is a newer field at present.
The technical issues that need to address：
1. solve how to combine RDF data body and OWL Horst rules build the pseudo- bilateral network of rule, wherein including pattern
The corresponding class node of triple and regular node, so as to be efficiently completed OWL in the case of extensive stream data
The reasoning of the whole rules of Horst.
2. combining the streaming scheme proposed proposes corresponding parallel inference scheme, so as to meet extensive stream data
The demand of distributed parallel reasoning.
The content of the invention
In order to solve the above problems, the present invention provides a kind of streaming RDF data parallel reasoning algorithm, for OWL Horst
Rule, with reference to the advantage of HAL algorithms, it is proposed that PRAS algorithms（Parallel Reasoning Algorithm for
Streaming RDF Data）.The algorithm can efficiently build in the case of extensive stream data and safeguard pseudo- Two-way Network
Network, and correct complete execution reasoning.
To achieve the above object, the present invention uses following technical scheme：A kind of streaming RDF data parallel reasoning algorithm, its
It is characterised by, comprises the following steps：S1：Loading rule node and pattern triple Pj_ RDD and Ok_ RDD is simultaneously saved in Redis collection
Group, builds the intermediate node midnode of link variable in rule, skips to S2；S2：Batch new data in timing reading data flow
The data itr_data that new_data and previous reasoning are produced；If it is pattern triple (Si,Pi,Oi), then skip to S3；If its
For example triple (si,pi,oi), then skip to S5；If it is empty that new_data, which is empty and itr_data, algorithm terminates；S3：If
Its corresponding class node Pj_ RDD or Ok_ RDD is present, then is referred to corresponding class node；If being not present, newly-built correspondence
Class node and be saved in Redis clusters；If its predicate belongs to Symmetric Property, S4 is skipped to；Otherwise S6 is skipped to；
Symmetric Property are the set for predicate in markers triple with symmetric relation.Symmetric properties triple
Set SymTriples is defined as follows：
；Wherein, Pj_ RDD is pattern triplet sets；S4：Data to input are sorted out and reasoning；S5：Produced for reasoning
Triple is stored and duplicate removal.
Compared with prior art, the present invention has advantages below：
1. combining OWL Horst rules and RDF ontology files constructing pseudo- bilateral network structure, the efficiency of streaming reasoning is improved.
2. combining the storage strategy of Redis clustering designs, the storage of duplicate removal and iterative data is carried out to triple, is reduced
The memory space and inference time of triple are repeated, so as to improve the efficiency of reasoning.
Brief description of the drawings
Fig. 1 is overall framework schematic diagram of the invention.
Fig. 2 is that pseudo- bilateral network builds figure.
Fig. 3 is loading rule and ontology data and builds pseudo- bilateral network.
Fig. 4 is OWL Horst rule relation figures.
Embodiment
Explanation is further explained to the present invention with specific embodiment below in conjunction with the accompanying drawings.
Streaming parallel inference proposed by the present invention is broadly divided into the pseudo- bilateral network of structure, stream data and sorted out and OWL
Three parts of reasoning of Horst rules.According to the characteristics of Spark Streaming and Redis, with reference to HAL algorithms and OWL
Horst rules and RDF data body, build the pseudo- bilateral network of rule, wherein including the corresponding class node of pattern triple
And regular node, set up intermediate node if the link variable that there is class in regular node；Then, timing acquisition Streaming
The data that batch new data in data flow and previous reasoning are produced as input data, the data to input sorted out or
Newly-built corresponding node is simultaneously stored to corresponding Redis clusters；Then, the triple data for input combine pseudo- bilateral network and sentenced
Whether the former piece that disconnected corresponding intermediate node or regular node are monitored all is met, and then the rule is made inferences, and is produced
Raw inference data.Finally, all data produced by deleting repetition inference data and this reasoning in real time are saved in Redis
As the input data of next reasoning in cluster, thus efficient realize the parallel streamings of RDF data OWL Horst rules
Reasoning.
Overall framework figure is referring to Fig. 1.
A kind of streaming RDF data parallel reasoning algorithm, it comprises the following steps：
S1：Loading rule node and pattern triple Pj_ RDD and Ok_ RDD is simultaneously saved in Redis clusters, builds in rule and connects
The intermediate node midnode of variable, skips to S2；
S2：The data itr_data that batch new data new_data and previous reasoning in timing reading data flow are produced；If its
For pattern triple (Si,Pi,Oi), then skip to S3；If it is example triple (si,pi,oi), then skip to S5；If new_data
It is empty for empty and itr_data, then algorithm terminates；
S3：If its corresponding class node Pj_ RDD or Ok_ RDD is present, then is referred to corresponding class node；If being not present,
Then newly-built corresponding class node and it is saved in Redis clusters；If its predicate belongs to Symmetric Property, S4 is skipped to；
Otherwise S6 is skipped to；Symmetric Property are the set for predicate in markers triple with symmetric relation.It is right
Attribute triplet sets SymTriples is claimed to be defined as follows：
；
Wherein, Pj_ RDD is pattern triplet sets；For example, OWL Horst rule in SymTriples=sameAs,
InverseOf, equivalentClass, equivalentProperty }；
S4：Data to input are sorted out and reasoning；
S5：Stored and duplicate removal for the triple that reasoning is produced.
Wherein S4 comprises the following steps：S41：If the triple data of input are pattern triple (Si,Pi,Oi), then will
The triple data of input are respectively with Pi+”_”+SiFor key, OiFor value and Pi+”_”+OiFor key, SiFor value, three are built
S in tupleiAnd OiBidirectional relationship, and be saved in Redis clusters, skip to S43；
S42：If the triple data of input are example triple (si,pi,oi), then the triple data of input are built<
si,(pi,oi)>、< pi , (si, oi)>With< oi , (si,pi)>Three key-value pairs, and Redis clusters are stored in, skip to
S43；
S43：The pseudo- bilateral network corresponding to new_data or itr_data is checked, and whether judges new_data or itr_data
The Rule monitored comprising regular node or intermediate nodem_ link_RDD, if the Rule that intermediate node is monitoredm_ link_RDD is then
S44 is skipped to, if the Rule that regular node is monitoredm_ link_RDD then skips to S45, otherwise skips to S2；Pseudo- bilateral network is referred to
To certain rule RuleiSet up regular node RuleiThe class being related in _ node, rule builds class node Classi_ node, such as
In the regular former piece of fruit intermediate node mid is then set up comprising link variablei_node；Regular RuleiLink variable refer to Rulei
In be used for the pattern triple that connects two former pieces, by the link variable information of each rule with<key,value>Shape
Formula is stored in Rulem_ link_RDD, wherein key store all pattern triples connected for former piece of the rule, value
Store the pattern triple of the rule conclusion part；The structure flow of pseudo- bilateral network is referring to Fig. 2.
S45：Judge the Rule monitoredmWhether _ link_RDD all meets, if then skipping to S46, otherwise skips to S2；
S46：Whether the corresponding all former pieces of judgment rule node all meet, if then the reasoning of executing rule produces ternary
Group, skips to S5；Otherwise S2 is skipped to.
S5 includes step in detail below：The triple produced for reasoning, is stored in entitled itr_data in Redis clusters
Set, and deduplication operation is carried out to the triple repeated, then regard itr_data set as reasoning input data next time
A part, S2 is skipped to if the order not stopped.
The PRAS algorithms of the present invention are according to the characteristics of Spark RDD and Redis clusters, with reference to the principle and OWL of HAL algorithms
Horst rules and RDF ontology datas, are built using the pseudo- bilateral network to rule, firstly for pattern triple
(Si,Pj,Ok) build corresponding class node Ok_ RDD or Pj_ RDD is simultaneously saved in Redis clusters, if P belongs to symmetric properties,
Bidirectional relationship is built to the S in the triple and O and Redis clusters are saved in.For all former pieces in quick judgment rule
Whether all meet, corresponding regular node is set up for strictly all rules, if containing link variable link_var in rule, build
Vertical intermediate node midnode, test condition information is stored in intermediate node and sets two-way between intermediate node and regular node
Communication；If connectionless variable, class node is joined directly together with regular node, test condition is stored in class node.To be advised in Fig. 2
Then exemplified by 8a, schematic diagram is as shown in Figure 3.By the heuristic information and the structure of symmetric properties between node, with reference to Redis collection
The efficient access of group, by required triple, the reading from Redis clusters in the way of inquiring about, reduces the reading of unrelated triple
Take and transmit, so as to improve overall Reasoning Efficiency.
The Map stages mainly complete data and sorted out and reasoning：If the batch fluxion in timing acquisition Streaming data flows
The data itr_data produced according to new_data or previous reasonings is ontology data, then is referred in corresponding class node, and more
The corresponding value of the node in new Redis clusters；If its attribute be symmetric properties, then respectively with " symm_ "+S and " symm_ "+O
For key, the bidirectional relationship of S and O in triple is built, and is stored in Redis clusters.If new_data or itr_data is real
Number of cases evidence, then to example triple (si,pi,oi), build< si, (pi,oi)>、< pi, (si,oi)>With<oi, (si,pi)
>Three key-value pairs, and it is stored in Redis clusters.Then the pseudo- bilateral network corresponding to new_data or itr_data is checked, and
Judge link variable or the corresponding all former pieces of regular node that the corresponding intermediate nodes of new_data or itr_data are monitored（Can
Multiple intermediate nodes can be included）Whether all meet, if then the reasoning of executing rule produces triple and is output to result
The Reduce stages；If part is met, the state of corresponding conditionses is changed.Set forth herein data sort out specifically walked with reasoning algorithm
It is rapid as follows：
Map phase algorithms
Input the triple that streaming triple data and previous reasoning are produced
Output<”new”,>
Step1 is for the triple data of input, (Si, Pj, Ok) ∈ SchemaTriple are referred to corresponding class node
And update Redis clusters；If PjFor symmetric properties, respectively with Pj+” _”+SiFor key, O is value and using P+ " _ "+O as key,
S is value, builds the bidirectional relationship of S and O in triple, and is stored in Redis clusters.Skip to Step3.
Step2 is for the triple data of input, (si,pj,ok) ∈ InstanceTriple, then to example triple (si,
pj,ok) build< si, (pj,ok) >、< pj, (si,ok) >With<ok, (si,pj) >Three key-value pairs are stored in Redis collection
Group.Skip to Step3.
Step3 checks (si,pj,ok) corresponding to pseudo- bilateral network, required data are read from Redis clusters, and judge
(si,pj,ok) corresponding the intermediate node link variable or the corresponding all former pieces of regular node monitored（May be comprising in multiple
Intermediate node）Whether all meet, if all met, skip to Step4.As fruit part is unsatisfactory for, then with reference to (Si,Pj,Ok) right
The monitoring information of intermediate node or class node is modified.
Step4 obtains the triple of reasoning generation according to the conclusion of current ruleAnd export<”new”,>。
With rule 8a and 8b in Fig. 4（inverseOf）Exemplified by, pseudo-code is described as follows：
Input: (S1, P1, O1)
Output: <”new”, >
Begin
If (S1, P1, O1) the ∈ SchemaTriple // triple be pattern triple, carry out classification preservation
{
If P1 equal “type”
sadd O1 S1
else {
sadd P1 (S1,O1)
If P1{/* predicates are that symmetric properties are to build to preserve subject S to ∈ SymmetriProperty1And O1Symmetrical pass
Be */
sadd P1+” _”+S1 O1
sadd P1+” _”+O1 S1
}
}
Else built when/* is example triple three key-value pairs preserve */
sadd S1 (P1,O1)
sadd P1 (S1,O1)
sadd O1 (S1,P1)
}
/ * reads inverseOf_S in Redis clusters1With inverseOf_O1Set to inverseOf*/
inverseOf smembers (“inverseOf_”+S1)
∪smembers (“inverseOf_”+O1)
If(inverseOf != null){
yield (“new”,( O1,P1, S1))
For (inverse in inverseOf.value){
yield (“new”,( O1, inverse, S1))
}
}
End
Assuming that containing pattern triple T (memberOf, owl in the batch flow data currently inputted:inverseOf,
) and example triple t (GraduateStudent0, memberOf, University0_Department0) member.It is first
First for pattern triple T, judge that inverseOf_RDD whether there is, if there is no then newly-built inverseOf_RDD and protect
(memberOf, member) is deposited into inverseOf_RDD；If there is being then saved directly to inverseOf_RDD.Then,
Due to inverseOf be symmetric properties, then respectively using inverseOf_memberOf as key, member be value and
InverseOf_member is key, and memberOf is value, and the bidirectional relationship for building memberOf and member is stored in
Redis clusters.For example triple t, build< GraduateStudent0, (memberOf, University0_
Department0)>、< memberOf, (GraduateStudent0, University0_Department0)>、<
University0_Department0, (GraduateStudent0, memberOf )>And it is saved in Redis clusters.Most
Afterwards, the set of inverseOf_ memberOf and inverseOf_ member in Redis clusters is read to inverseOf, time
Go through inverseOf and export (GraduateStudent0, member, University0_Department0).
, can by the structure and the bidirectional relationship of storage of symmetric properties similar to regular 8a and 8b containing symmetric properties
Quickly to find out the triple of correlation in Redis clusters, so as to improve Reasoning Efficiency.
With rule 15 in Fig. 4（someValuesFrom）Exemplified by, pseudo-code is described as follows：
Input: (S1, P1, O1)
Output: <”new”, >
Begin
If (S1, P1, O1) { // triple is pattern triple to ∈ SchemaTriple, carries out classification preservation
If P1 equal “type”
sadd O1 S1
else {
sadd P1 (S1,O1)
If P1{ // predicate is that symmetric properties are to build the symmetric relation for preserving subject S1 and O1 to ∈ SymmetriProperty
sadd P1+” _”+S1 O1
sadd P1+” _”+O1 S1
}
}
Else // build three key-value pairs when being example triple and preserve
sadd S1 (P1,O1)
sadd P1 (S1,O1)
sadd O1 (S1,P1)
}
someValuesFrom_set Smembers (" someValuesFrom ")/* is read in Redis clusters
SomeValuesFrom set */
onProperty_set smembers (“onProperty”)
For (svf in someValuesFrom_set) {
For (op in onProperty_set) {
If(svf.v equals op.v){
temp_w smembers (svf.w)
x_type_w= temp_w.filter(x =>X.p==" type ")/* filters out the ternary that p in temp_w is type
Group */
u_p_x smembers (op.p)
result = u_p_x.filter(t=>
t.x==x_type_w.x
Yield (" new ", (t.u, type, svf.v))) former piece in rule is attached by/*, produce the reasoning results */
}
}
}
End
Assuming that containing pattern triple T1 (Chair, owl in the batch flow data currently inputted:someValuesFrom,
Department), T2 (Chair, owl:OnProperty, headOf) and example triple t1 (FullProfessor7,
HeadOf, University0_Department0), t2 (University0_Department0, rdf:type,
Departmment).Firstly for pattern triple T1 and T2, someValuesFrom_RDD and onProperty_RDD are judged
It whether there is, preserve if there is no then newly-built someValuesFrom_RDD and onProperty_RDD and respectively (Chair,
Department) to someValuesFrom_RDD and preservation (Chair, headOf) into onProperty_RDD；If deposited
Then it is being saved directly to someValuesFrom_RDD or onProperty_RDD.For example triple t1, build<
FullProfessor7, (headOf, University0_Department0)>、< headOf, (FullProfessor7,
University0_Department0)>、< University0_Department0 , (FullProfessor7,
headOf)>And it is saved in Redis clusters, the similar aforesaid operations of t2.Then, read respectively in Redis clusters
SomeValuesFrom and onProperty set is traveled through to someValuesFrom_set and onProperty_set
SomeValuesFrom_set and onProperty_set, now the Chair in someValuesFrom_set with
OnProperty_set Chair is identical, is then respectively that key obtains Redis clusters by key and headOf of Department
In two set；Finally FullProfessor7 is connected with Chair and (FullProfessor7, rdf is exported:type,
Chair)。
Similar to the rule 15 of multi-link variable, for key of the pattern triple by class node, can quickly from
Obtained in Redis clusters；For the example triple of association, using the storage strategy of example triple in Redis, pass through connection
The value of variable finds out the example triple of correlation, so as to improve Reasoning Efficiency.
The Reduce stages mainly preserve to the data that reasoning is produced.The triple produced for reasoning, is stored in
It is entitled in Redis clusters " itr_data " set, and deduplication operation is carried out to the triple repeated, then will " itr_
Data " gathers the part as next reasoning input data.Set forth herein data deduplication and storage algorithm specific steps such as
Under：
Reduce algorithms
Input<”new”, Iterator<String> values>
Export null
Step1. the SchemaTriple and InstanceTriple of input are stored in using itr_data as set name
In Redis clusters, the reading for next reasoning.
For definitely duplicate removal and storage of the Reduce stages to input data, pseudo-code is described as follows：
Input: <”new”, Iterator <String> values>
Output: null
Begin:
del itr_data
itr for each values
Value in sadd itr_data itr.value/* traversals values is added to the itr_data set of Redis clusters
Middle */
End
Can be obtained by above-mentioned pseudo-code, in the Reduce stages, by the triple of input by Redis set carry out duplicate removal and
Storage, is the preparation that data are carried out in next reasoning.
Algorithm complexity analysis is the important indicator for weighing an efficiency of algorithm, and the complexity of PRAS algorithms of the invention is divided
Analysis has different modes from centralized algorithm.Analyze PRAS algorithms complexity when, can be broken down into Map and
Two stages of Reduce carry out algorithm complexity analysis.If experimental data, which is concentrated, includes N number of triple, Redis data are read
Time be set to Map tasks during t, and MapReduce and line number is set to k, the example triple number that the Reduce stages receive
Be set to m, Reduce tasks and line number is set to x.Because PRAS algorithms are in triple of the Map stages to each input, with reference to class
Node or intermediate node run-down, you can judge whether the triple can participate in some rule-based reasonings, if participating in follow-up rule
Then reasoning, then by reading the former piece data in Redis and then reasoning obtains the reasoning results.Therefore, the time in Map stages is complicated
Property is：O(t*N/k).The triple of each input is sorted out in the Reduce stages, therefore, the time in Reduce stages answers
Polygamy is：O(m/x).
Above is presently preferred embodiments of the present invention, all changes made according to technical solution of the present invention, produced function is made
During with scope without departing from technical solution of the present invention, protection scope of the present invention is belonged to.
Claims (3)
1. a kind of streaming RDF data parallel reasoning algorithm, it is characterised in that comprise the following steps：
S1：Loading rule node and pattern triple Pj_ RDD and Ok_ RDD is simultaneously saved in Redis clusters, builds in rule and connects
The intermediate node midnode of variable, skips to S2；
S2：The data itr_data that batch new data new_data and previous reasoning in timing reading data flow are produced；If its
For pattern triple (Si,Pi,Oi), then skip to S3；If it is example triple (si,pi,oi), then skip to S5；If new_data
It is empty for empty and itr_data, then algorithm terminates；
S3：If its corresponding class node Pj_ RDD or Ok_ RDD is present, then is referred to corresponding class node；If being not present,
Newly-built corresponding class node is simultaneously saved in Redis clusters；If its predicate belongs to Symmetric Property, S4 is skipped to；It is no
Then skip to S6；Symmetric Property are the set for predicate in markers triple with symmetric relation.Symmetrically
Attribute triplet sets SymTriples is defined as follows：
；
Wherein, Pj_ RDD is pattern triplet sets；
S4：Data to input are sorted out and reasoning；
S5：Stored and duplicate removal for the triple that reasoning is produced.
2. a kind of streaming RDF data parallel reasoning algorithm according to claim 1, it is characterised in that：S4 includes following step
Suddenly：S41：If the triple data of input are pattern triple (Si,Pi,Oi), then by the triple data of input respectively with Pi
+”_”+SiFor key, OiFor value and Pi+”_”+OiFor key, SiFor value, S in triple is builtiAnd OiBidirectional relationship,
And Redis clusters are saved in, skip to S43；
S42：If the triple data of input are example triple (si,pi,oi), then the triple data of input are built<
si,(pi,oi)>、< pi , (si, oi)>With< oi , (si,pi)>Three key-value pairs, and Redis clusters are stored in, skip to
S43；
S43：The pseudo- bilateral network corresponding to new_data or itr_data is checked, and whether judges new_data or itr_data
The Rule monitored comprising regular node or intermediate nodem_ link_RDD, if the Rule that intermediate node is monitoredm_ link_RDD is then
S44 is skipped to, if the Rule that regular node is monitoredm_ link_RDD then skips to S45, otherwise skips to S2；Pseudo- bilateral network is referred to
To certain rule RuleiSet up regular node RuleiThe class being related in _ node, rule builds class node Classi_ node, such as
In the regular former piece of fruit intermediate node mid is then set up comprising link variablei_node；Regular RuleiLink variable refer to Rulei
In be used for the pattern triple that connects two former pieces, by the link variable information of each rule with<key,value>Shape
Formula is stored in Rulem_ link_RDD, wherein key store all pattern triples connected for former piece of the rule, value
Store the pattern triple of the rule conclusion part；
S45：Judge the Rule monitoredmWhether _ link_RDD all meets, if then skipping to S46, otherwise skips to S2；
S46：Whether the corresponding all former pieces of judgment rule node all meet, if then the reasoning of executing rule produces ternary
Group, skips to S5；Otherwise S2 is skipped to.
3. a kind of streaming RDF data parallel reasoning algorithm according to claim 1, it is characterised in that：S5 includes following tool
Body step：The triple produced for reasoning, is stored in the set of entitled itr_data in Redis clusters, and to repetition
Triple carries out deduplication operation, then itr_data is gathered to the part as next reasoning input data, if do not stopped
Order only then skips to S2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710246309.2A CN106980901B (en) | 2017-04-15 | 2017-04-15 | Streaming RDF data parallel reasoning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710246309.2A CN106980901B (en) | 2017-04-15 | 2017-04-15 | Streaming RDF data parallel reasoning algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106980901A true CN106980901A (en) | 2017-07-25 |
| CN106980901B CN106980901B (en) | 2019-09-13 |
Family
ID=59346065
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710246309.2A Active CN106980901B (en) | 2017-04-15 | 2017-04-15 | Streaming RDF data parallel reasoning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106980901B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108763451A (en) * | 2018-05-28 | 2018-11-06 | 福州大学 | Streaming RDF data parallel reasoning algorithm based on Spark Streaming |
| CN108875953A (en) * | 2018-06-08 | 2018-11-23 | 福州大学 | A kind of complex rule reasoning design method extending DL operator |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105468702A (en) * | 2015-11-18 | 2016-04-06 | 中国科学院计算机网络信息中心 | Large-scale RDF data association path discovery method |
| CN105808853A (en) * | 2016-03-09 | 2016-07-27 | 哈尔滨工程大学 | Engineering application oriented body establishment management and body data automatic obtaining method |
-
2017
- 2017-04-15 CN CN201710246309.2A patent/CN106980901B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105468702A (en) * | 2015-11-18 | 2016-04-06 | 中国科学院计算机网络信息中心 | Large-scale RDF data association path discovery method |
| CN105808853A (en) * | 2016-03-09 | 2016-07-27 | 哈尔滨工程大学 | Engineering application oriented body establishment management and body data automatic obtaining method |
Non-Patent Citations (4)
| Title |
|---|
| SANKET CHINTAPALLI ETC: "Benchmarking Streaming Computation Engines:Storm, Flink and Spark Streaming", 《2016 IEEE INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS》 * |
| 李璞等: "一种面向结构化数据源的语义标注和挖掘方法", 《南阳师范学院学报》 * |
| 汪璟玢等: "结合RATe的RDF数据分布式并行推理算法", 《模式识别与人工智能》 * |
| 董书暕等: "ＨＳＳＴ＋基于分布式内存数据库的", 《计算机科学》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108763451A (en) * | 2018-05-28 | 2018-11-06 | 福州大学 | Streaming RDF data parallel reasoning algorithm based on Spark Streaming |
| CN108763451B (en) * | 2018-05-28 | 2022-03-11 | 福州大学 | Streaming RDF data parallel reasoning algorithm based on Spark Streaming |
| CN108875953A (en) * | 2018-06-08 | 2018-11-23 | 福州大学 | A kind of complex rule reasoning design method extending DL operator |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106980901B (en) | 2019-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bo et al. | Beyond low-frequency information in graph convolutional networks | |
| Wu et al. | Graph summarization for attributed graphs | |
| CN105912721B (en) | RDF data distributed semantic parallel inference method | |
| Chatzianastasis et al. | Graph ordering attention networks | |
| Chen et al. | Learning data-driven drug-target-disease interaction via neural tensor network | |
| Singh et al. | Performance Measure of Similis and FPGrowth Algo rithm | |
| CN106980901A (en) | Streaming RDF data parallel reasoning algorithm | |
| Caragea et al. | Analysis and synthesis of agents that learn from distributed dynamic data sources | |
| CN107016110A (en) | With reference to the OWLHorst regular distribution formula parallel reasoning algorithms of Spark platforms | |
| Zhou et al. | Multi-channel graph convolutional networks | |
| Yan et al. | Osgnn: Original graph and subgraph aggregated graph neural network | |
| US20190347302A1 (en) | Device, system, and method for determining content relevance through ranked indexes | |
| Balaji et al. | Distributed graph path queries using spark | |
| Hong et al. | Using divide-and-conquer GA strategy in fuzzy data mining | |
| Wang et al. | Binary Graph Convolutional Network With Capacity Exploration | |
| Dong et al. | An improved niching binary particle swarm optimization for feature selection | |
| Ding et al. | Efficient probabilistic skyline query processing in mapreduce | |
| Freschi et al. | Multiobjective optimization and artificial immune systems: a review | |
| Freitas et al. | On the visualization of trade-offs and reducibility in many-objective optimization | |
| Wan et al. | Tackling Over-Smoothing: Graph Hollow Convolution Network with Topological Layer Fusion | |
| CN108763451B (en) | Streaming RDF data parallel reasoning algorithm based on Spark Streaming | |
| Zhang et al. | P-EAARM: A generic framework based on spark for eas-based association rule mining | |
| Choudhary et al. | PersonaSAGE: A Multi-Persona Graph Neural Network | |
| Zhai et al. | A three-stage method for classification of binary imbalanced big data | |
| Zhao et al. | Multi-View Heterogeneous Graph Attention Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |