OA17269A - Index constructing method, search method, device and system. - Google Patents

Abstract

The present disclosure discloses an index constructing method, search method, device and system. The index constructing method includes : constructing data obtained in each preset time period into a primary index segment, respectively, and merging the constructed primary index segments into a primary index segment set after each preset time period is over; copying and merging each primary index segment in the primary index segment set into a corresponding large index segment, to obtain a large index segment set; and sending the primary index segments in the primary index segment set or the large index segment in the large index segment set ; and sending the primary index segments set or the large index segments in the large index segment set to a search device to provide a search service. The present disclosure constructs the primary index segments on the basis of the, merges the constructed primary index segments into large index segments, and provides the search service simultaneously with the primary index segment or the large index segments, thus, improving search speed.

Description

The présent disclosure relates to search technology, and more particularly to, an index constructing method, search method, device and system.

Backqround

With the popularity of the network and the increasing amount of information, application ranges of search technology are more and more widely. In orderto achieve searching, it needs to construct data into indexes in advance and then perform searching according to the indexes constructed in advance. Thus, a manner of constructing indexes can détermine search speed.

In the related art, one manner of constructing indexes is to establish two index segments including one large index segment and one small index segment according to file size, merge new data into the small index segment in real time, and merge at some time the large index segment and the small index segment into a new large index segment. During searching, the large index segment and the small index segment are simultaneously searched, and then results of the searching in both .of the large index segment and the small index segment are merged.

In the process of implémentation of the présent disclosure, inventors find that the related art has at least the following disadvantages:

Since in the related art, it is needed to merge the new data and the small index segment and then merge the small index segment and the large index segment before providing search service, thus, the search speed is affected.

Summary

In order to improve search speed, one example of the présent disclosure provides an index constructing method, search method, device and system. The technical solution is as follows:

On one hand, an index constructing method is provided, the methods includes:

constructing data obtained in each preset time period into a primary index segment, respectively, and merging the constructed primary index segment into a primary index segment set after each preset time period is over;

i copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, to obtain a large index segment set; and sendîng the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device to provide search service.

An index constructing device is also provided and includes:

a constructing module configured to construct data obtained in each preset time period into a primary index segment, respectively, and merge the constructed primary index segment into a primary index segment set after each preset time period is over;

a first merging module configured to copy each primary index segment in the primary index segment set obtained by the constructing module and merge the copied primary index segments into a corresponding large index segment to obtain a large index segment set; and a sendîng module configured to send the primary index segments in the primary index segment set obtained by the constructing module or the large index segments in the large index segment set obtained by the first merging module to a search device to provide search service.

On the other hand, a search method is also provided and includes:

every each preset period, obtaining a first-level index segment from an index constructing device, and merging the obtained first-level index segment into a first-level index segment set;

copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

providing search service according to the first-level index segments in the first-level index segment set and the second-level index segments in the second-level index segment set.

A search device is also provided and includes:

an obtaining module configured to, every each preset period, obtain a first-level index segment from an index constructing device, and merge the obtained first-level index segment into a first-level index segment set;

a second merging module configured to copy each first-level index segment in the first-level index segment set obtained by the obtaining module, and merge the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

a search module configured to provide search service according to the first-level index segments in the first-level index segment set obtained by the obtaining module and the second-level index segments in the second-level index segment set obtained by the second merging module.

On the other hand, a search system is also provided and includes an index constructing device and more than one search devices;

the index constructing device is the above index constructing device;

the search device is the above search device.

Bénéficiai effects of the technical solution provided in embodiments of the présent disclosure include:

As mentioned above, since in the related art, it is needed to merge the new data and the small index segment and then merge the small index segment and the large index segment before providing search service, thus, the search speed is affected; while after adopting the technical solution provided in embodiments ofthe présent disclosure, constructing primary index segments on the basis of time, merging the constructed primary index segments into the large index segment, and provides the search service simultaneously with the primary index segments or the large index segments, i.e., the already constructed primary index segments or large index segments can be used to provide search service at the time of constructing indexes, one thread for constructing indexes and one thread for searching are executed in parallel, thereby improving the search speed.

Brief Description of Drawings

Exemplary examples of présent disclosure will be described in detail with référencé to the accompanying drawings to make one skilled in the art clearer of the above and other features and advantages ofthe présent disclosure. In drawings,

Fig. 1 is a flow chart of an index constructing method according to a first example ofthe présent disclosure;

Fig. 2 is a flow chart of an index constructing method according to a second example of the présent disclosure;

Fig. 3 is a schematic diagram of merging index segments according to the second example of the présent disclosure;

Fig. 4 is a schematic diagram showing a process of constructing index according to the second example ofthe présent disclosure;

Fig. 5 is a first schematic diagram of an index constructing device according to a third example of the présent disclosure;

Fig. 6 is a second schematic diagram of the index constructing device according to the third example of the présent disclosure;

Fig. 7 is a third schematic diagram of the index constructing device according to the third example of the présent disclosure;

Fig. 8 is a schematic diagram of a sending module according to the third example of the présent disclosure;

Fig. 9 is a flow chart of a search method according to a forth example of the présent disclosure;

Fig. 10 is a flow chart of a search method according to a fîfth example of the présent disclosure;

Fig. 11 is a schematic diagram of searching according to the fifth example of the présent disclosure;

Fig. 12 is a schematic diagram of a search device according to a sixth example of the présent disclosure;

Fig. 13 is a schematic diagram of an obtaining module according to the sixth example of the présent disclosure;

Fig. 14 is a schematic diagram of a search system according to a seventh example of the présent disclosure; and

Fig. 15 is a schematic diagram of another search system module according to the seventh example of the présent disclosure.

Detailed Description

In order to make the objectives, technical solutions and advantages of the présent disclosure more clear, various examples of the présent disclosure will be described in detail with reference to the accompanying drawings.

First example

This example provides an index constructing method. Referring to Fig. 1, a flow of the method provided in this example is specifically as follows:

101: constructing data obtained in each preset time period into a primary index segment, respectively, and merging the constructed primary index segment into a primary index segment set after each preset time period is over;

102: copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, to obtain a large index segment set;

103: sending the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device so as to provide search service.

In order to clearly illustrate the method provided in this example, the method provided in this example will be further explained hereinafter with reference to spécifie content of a second example. Details refer to the content ofthe following second example.

Second example

This example provides an index constructing method. Referring to Fig. 2, a flow of the method provided in this example is specifically as follows:

201: constructing data obtained in each preset time period into a primary index segment, respectively.

For this step, this example does not limit size of the preset time period. For convenience of description, this example only takes the size of the preset time period being 5 minutes as an example, and then the method provided in this example constructs data obtained in every 5 minutes into a primary index segment, respectively. Since constructing data into index segments is already very mature existing technology, thus, the manner of constructing the obtained data into a primary index segment is not specifically described and limited here in this example.

Specifically, the constructed primary index segment contains multiple independent index structures of data, and the independent index structures include but not limited to an inverted index, a delete table, document basic information and customized other information. Each item in an index table of the inverted index includes an attribute value and an address of each record having the attribute value. Not by the records to détermine attribute value, but by the attribute value to déterminethe location ofthe records, which is called the inverted index. Index objectsofthe inverted index are words, etc., in a document or a set of documents. The inverted index is used to store storage locations of words in a document or a set of documents, and is a common index mechanism of a document or a set of documents. Each document has its own separate document id, and thus different documents can be able to be distinguished. With respect to the spécifie content of the document basic information, it is not specifically limited in this example. In this example, in order to use information already existing in the index segments to construct appropriate and fast memory data structure, one way of initializing the constructed primary index segments is taken in this example, and the data structure of the primary index segments after initialization can be shown in following Table 1 :

Table 1 segment header:

location and length of mapping table from key word to inverted index location and length of inverted index location and length of delete table location and length of mapping table from document id to document basic information location and length of document basic information other information mapping table from key word to inverted index inverted index delete table mapping table from document id to document basic information document basic information

In table 1, whether the mapping table from key word to inverted index or the mapping table from document id to document basic information, each mapping table is composed of a hash array and a collision list, and the hash array and the collision list form a close array structure. During initialization, data can be directly taken as a mapping table.

202: after each preset time period is over, merging the constructed primary index segments into 10 a primary index segment set.

For this step, still taking the preset time period being 5 minutes as an example, then in every 5 minutes, ail obtained data can be constructed into a primary index segment; after reaching 5 minutes, the primary index segment is merged into a primary index segment set. The primary index segment set includes primary index segments constructed in a plurality of 5 minutes.

Preferably, since the method provided in this example will merge the primary index segments in the primary index segment set into a corresponding large index segment set, thus, in order to save memory resources, after merging the constructed primary index segments into a primary index segment set, the method provided in this example further includes:

every each preset period, deleting primary index segments which are constructed before the each preset period from the primary index segment set.

This example does not limit spécifie size of the preset period, so long as it can be ensured that in each preset period, primary index segments to be deleted hâve been already merged into a large index segment set.

203: copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, to obtain a large index segment set.

Specifically, Fig. 3 shows a whole process of constructing the primary index segment, merging the primary index segment into the primary index segment set, and copying and merging each primary index segment in the primary index segment set into the corresponding large index segment.

When copying each primary index segment in the primary index segment set and merging the copied primary index segments into the corresponding large index segment, each primary index segment in the primary index segment set may be copied in tum according to time sequence, and the copied primary index segments are merged into the corresponding large index segment. The corresponding large index segment is a large index segment which is in the large index segment set and neighbors a free memory.

In order to facilitate understanding, in this example, the merging manner of this step will be described in details with reference to a merging process shown in Fig. 4. In spécifie realization, the method provided in this example may allocate in advance a continuous chunk of memory, and the memory stores a plurality of large index segments and a free memory area. Every time selecting a large index segment neighboring the free memory area for merging, one large index segment neighboring and locating above the free memory area may be selected, and one large index segment neighboring and locating below the free memory area may also be selected. If there is no large index segment above and below the free memory area, the primary index segments are directly merged in the free memory. Referring to Fig. 4a, in this example, as an instance, one large index segment, which neighbors the free memory area, locates above the free memory area and is selected from the large index segment set, is the large index segment 4. One primary index segment needed to be merged will be merged into the corresponding large index segment 4. After selecting to merge the primary index segment into the corresponding large index segment 4, one new large index segment 4 is obtained through mergence. ln subséquent mergence of primary index segments, referring to Fig. 4(b), according to the last sélection manner, one selected large index segment which neighbors the free memory area and locates above the free memory area, is the large index segment 3, the obtained large index segment set after mergence is shown in Fig. 4(b).

Since when merging the primary index segments into the correspondîng large index segment, the way of merging the primary index segments into a large index segment which neighbors the free memory is adopted, combined with Fig. 3 and the above discussed contents, it can be seen that the merging manner provided in this example can ensure a continuous free memory, i.e., there is no void in the memory, thereby improving utilization of the memory. Further, in the merging manner provided in this example, there is no intersection between a large index segment and a primary index segment which are needed to be merged; thus, before the completion of the merging, this large index segment and the primary index segment can still continue providing search service.

Preferably, after copying each primary index segment in the primary index segment set and merging the copied primary index segments into a correspondîng large index segment, to obtain a large index segment set, the method provided in this example further includes: backing up each large index segment in the large index segment set. With respect to spécifie backup mode, this example does not place any make spécifie limits on it. ln actual applications, each large index segment in the obtained large index segment set may be backed up to a local disk.

204: sending the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device to provide search service.

For this step, when constructing an index segment in the method provided in this example, the constructed index segment at least further includes a seqno (sequence number) value and an IsComplete sign of whether being complété in addition to the common inverted index, delete table, document basic information and other information. The seqno value is configured to distinguish constructed different index segments. This example does not limit size of the sequence number value. Taking the seqno value being a 64-bit integer as an example, for each constructed primary index segment, the seqno value of each constructed primary index segment strictly increases progressively starting from 1. This example does not limit rate of increasing progressively. As an instance, the seqno value can be progressively increased by adding 1 every time. Every time a new primary index segment is constructed, a new seqno value is allocated to the new primary index segment. When a large index segment and a primary index segment are merged into a new large index segment, a sequno value of the new large index segment is equai to the larger of a sequno value ofthe large index segment and a sequno value ofthe primary index segment. The IsComplete sign refers to whether one index segment is complété. In this example, for one primary index segment, after the primary index segment has been merged into a primary index segment set, this primary index segment is complété, i.e., this primary index segment includes ail data within 5 minutes. If the preset time period is not reached, one primary index segment may continue to obtain data subsequently, which means that the currently constructed primary index segment does not include ail data within 5 minutes; thus, this primary index segment is not complété. While for each large index segment, since each large index segment is obtained through merging complété primary index segments, thus, each large index segment is complété. Preferably, in order to check index data, one index segment may further include an md5 check value. The md5 check value refers to md5 values of ail bytes of the index segment.

In combination with the above information contained in the primary segment, when sending the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device to provide search service, the method provided in this example specifically includes:

receiving an index segment application sent from the search device, the application carrying a sequence number and a sign of whether being complété;

sending corresponding index segments to the search device according to the sequence number and sign of whether being complété carried in the application, the corresponding index segments being primary index segments of the primary index segment set or large index segments of the large index segment set.

The purpose why the search device sends the index application is to obtain newly constructed index segments so as to update index segments already in the search device itself, thus, the sequence number value carried in the application sent from the search device is the largest sequence number value of those ofthe index segments already in the search device. For situations in which the search device has no index segment due to reasons such as initial start-up, restart, failure, when the search device sends the index segment application, the sequence number value carried in the application is set to null. The situation in which the sequence number value is null will be described in details when introducing a searching method as shown in step 1001 ofthe example five.

Further, returning corresponding index segments to the search device according to the sequence number and sign of whether being complété carried in the application specifically includes:

if the sequence number value carried in the application is null, reading and sending backed up index segments to the search device;

or, if the sequence number value carried in the application is non-null and the application carnes a sign of being incomplète, retuming an index segment with a sequence number value the same as the sequence number value carried in the application to the search device;

or, if the sequence number value carried in the application is non-null and the application carries a sign of being complété, retuming an index segment with a sequence number value which is the next sequence number value of the sequence number value carried in the application to the search device.

In the method provided in this example, through constructing a plurality of primary index segments on the basis of time, merging the constructed primary index segments into the large index segment, and providing the search service simultaneously with the primary index segments or the large index segments, the search speed can be improved. When merging index segments, the index segments are preferably merged into an index segment neighboring the free memory area, thereby improving utilization of the memory. Further, when constructing a primary index segment, existing data in the index segment can be used to initialize the index segment so as to construct appropriate and fast memory data structures, thus, the search speed can be further improved.

Third example

This example provides an index constructing device, which can be used to perform the index constructing method ofthe above second example. Referring to Fig. 5, the index constructing device includes:

a constructing module 501 configured to construct data obtained in each preset time period into a primary index segment, respectively; and merge the constructed primary index segment into a primary index segment set after each preset time period is over;

a first merging module 502 configured to copy each primary index segment in the primary index segment set obtained by the constructing module 501 and merge the copied primary index segments into a corresponding large index segment to obtain a large index segment set;

a sending module 503 configured to send the primary index segments in the primary index segment set obtained by the constructing module 501 or the large index segments in the large index segment set obtained by the first merging module 502 to a search device to provide search service.

Specifically, the first merging module 502 is specifically configured to copy, in tum according to time sequence, each primary index segment in the primary index segment set obtained by the constructing module 501, and merge the copied primary index segments into a corresponding large index segment which is a large index segment in the large index segment set and neighbors a free memory.

Preferably, referring to Fig. 6, the device further inciudes:

a délétion module 504 configured to, after the constructing module 501 merges the constructed primary index segment into the primary index segment set, every each preset period, delete primary index segments which are constructed before the each preset period from the primary index segment set.

Preferably, referring to Fig. 7, the device further inciudes:

a backup module 505 configured to, after the first merging module 502 copies each primary index segment in the primary index segment set obtained by the constructing module 501 and merges the copied primary index segments into a corresponding large index segment to obtain a large index segment set, back up each large index segment in the large index segment set.

Each of the primary index segment constructed by the constructing module 501 and the large index segment obtained by the first merging module 502 through merging inciudes a sequence number value and a sign of whether being complété;

accordingly, when the first merging module 502 copies each primary index segment in the primary index segment set obtained by the constructing module 501 and merges the copied primary index segments into a corresponding large index segment, a sequence number value of the large index segment after merged is the larger of a sequence number value of the primary index segment and a sequence number value of the large index segment when merging;

accordingly, referring to Fig. 8, the sending module 503 specifically inciudes:

a first receiving unit 503a configured to receive an index segment application sent from the search device, the application carrying a sequence number value and a sign of whether being complété;

a first sending unit 503b configured to, send corresponding index segments to the search device according to the sequence number value and sign of whether being complété carried in the application received by the first receiving unit 503a, the corresponding index segments being primary index segments of the primary index segment set or large index segments of the large index segment set.

Further, the first sending unit 503b is specifically configured to, if the sequence number value carried in the application received by the first receiving unit 503a is null, read and send backed up index segments to the search device; or, if the sequence number value carried in the application received by the first receiving unit 503a is non-null and the application cames a sign of being n

incomplète, retum an index segment with a sequence number value the same as the sequence number value carried in the application to the search device; or, if the sequence number value carried in the application received by the first receiving unit 503a is non-null and the application carries a sign of being complété, retum an index segment with a sequence number value which is the next sequence number value of the sequence number value carried in the application to the search device.

ln the device provided in this example, through constructing a plurality of primary index segments on the basis of time, merging the constructed primary index segments into the large index segment, and providing the search service simultaneously with the primary index segments or the large index segments, thus, the search speed can be improved. Further, when merging index segments, the index segments are preferably merged into an index segment neighboring the free memory area, thereby improving utilization of the memory.

Example four

This example provides a search method, referrîng to Fig. 9, a flow chart of the search method provided in this example is specifically as foilows:

901: every each preset period, obtaining a first-level index segment from an index constructing device, and merging the obtained first-level index segment into a first-level index segment set;

902: copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

903: providing search service according to the first-level index segments in the first-level index segment set and the second-level index segments in the second-level index segment set.

ln order to better explain the method provided in this example, the method provided in this example will be described in detail hereinafter with reference to spécifie contents of example five, referrîng to the contents of the following example five.

Example five

This example provides a search method, referrîng to Fig. 10, a flow chart of the search method provided in this example is specifically as foilows:

1001: every each preset period, sending an index segment application to an index constructing device, the application carrying a sequence number value and a sign of whether being complété.

For this step, in order to update in real time already existed index segments so as to ensure accuracy of search results, it is needed to continue obtaining index segments from the index constructing device, thus, the preset period can be set to be a few seconds or minutes. This example does specifically limit size ofthe preset period.

Further, in order to be able to obtain the needed index segments from the index constructing device, every each preset period, the search device sends an index segment application to the index constructing device, the application each cames a sequence number value and a sign of whether being complété. In order to obtain new index segments constructed by the index constructing device so as to update index segments used for searching, when sending the index segment application in this step, the sequence number value carried in the application is the largest sequence number value of those ofthe index segments already existed in the search device.

For situations in which the search device has no index segment due to reasons such as initial start-up, restart, failure, when sending the index segment application in this step, the sequence number value carried in the application is determined to be null. The determining manner can be processed separately according to the following two different situations:

First situation: the search device has no index segment after the search device is restarted due to various reasons (such as initial deployment, crash, and power outage).

For this situation, when the search device sends the index segment application to the index constructing device, the seqno value carried in the application is set to null. After the index constructing device receives the index segment application whose sequence number value is null, the index constructing device retums the most recent backup data read from a disk. As an instance, the index constructing device backs up the large index segments once every one hour and deletes primary index segments in the primary index segment set every two hours; then, the next time applying for index segments, the new index segments are in the primary index segment set.

The second situation: when network environment is out of state (for example, the network cannot be used in continuous 30 minutes), the search device cannot communicate with the index constructing device in a long-term, there is a large différence between index segments used by the search device during searching and index segments constructed by the index constructing device.

For this situation, as an instance, the index constructing device backs up the large index segments once every one hour and deletes primary index segments in the primary index segment set every two hours. When sending the index segment application to the index constructing device in this step, if the sequence number value carried in the application still is the largest sequence number value of those of the index segments already existed in the search device, then, after the search index constructing device receives the sequence number value carried in the application: if the search device finds an index segment corresponding to the sequence number value and détermines that a time gap in construction of the index segment corresponding to the sequence number value is within two hours (Le., off the network within two hours), the normal process is still performed (because the first-level index segment set of the search device contains first-level index segments of two hours); if determining that a time gap in construction of the index segment corresponding to the sequence number value is greater than two hours (i.e., off the network more than two hours), then the index constructing device can issue an error notification to the search device, so as to notifÿ the search device to empty the current index segments and perform according to the processing manner of the first situation. This example does not limit the manner in which the index constructing device sends the notification message to the search device. For this situation, other processing manners may be adopted, and this example does not specifically limit it so long as ensuring that the search device is able to obtain index segments constructed by the index constructing device after the search device sends failure.

It can be seen from the above example two, since each index segment includes a sequence number value and a sign of whether being complété, thus, when the index constructing device receives an application carrying a sequence number value and a sign of whether being complété, the index constructing device can détermine to retum which index segment. The manner in which the index constructing device returns index segments refers to related description in the step 201 of the above example two, and is not repeated here.

1002: receiving an index segment sent from the index constructing device according to the sequence number value and the sign of whether being complété carried in the application, taking the received index segment as a first-level index segment and merging the first-level index segment into a first-level index segment set.

For this step, in actual implémentation, when merging a received first-level index segment into a first-level index segment set, the received first-level index segment can be arranged in the first-level index segment set in order according to the received sequence. If a received first-level index segment is an incomplète index segment, this received first-level index segment is directly arranged as the last index segment in the first-level index segment set. If a received first-level index segment is a complété index segment, this received first-level index segment covers the last index segment in the first-level index segment set. About how to détermine whether the received first-level index segment is complété, it can be obtained through the sign of whether being complété included in the first-level index segment.

It should be noted, the above steps 1001 and 1002 are examples for illustrating the step 901 of example four. In actual applications, other manners can be adopted to achieve obtaining a first-level index segment from an index constructing device every each preset period of the step 901. This example does not specifically limit this.

1003: copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set.

For this step, one process of merging the first-level index segments and the second-level index segments can be shown in Fig. 11. The merging principle is the same as the principle of merging the primary index segments and the large index segment in the step 203 of example two, i.e., each first-level index segment in the first-level index segment set may be copied in tum according to time sequence and the copied first-level index segments are merged into the corresponding second-level index segment. The corresponding second-level index segment is a second-level index segment which is in the second-level index segment set and neighbors a free memory. A sequence number value of the second-level index segment after merged is the larger sequence number value of a sequence number value of the first-level index segment and a sequence number value of the second-level index segment when merging.

1004: providing search service according to the first-level index segments in the first-level index segment set and the second-level index segments in the second-level index segment set.

For the search method provided in this example, in actual applications, the time of merging a complété primary index segment constructed in one preset time period and a large index segment is less than the preset time period, thus, searching and constructing indexes are fully synchronized. A formula for calculating a total delay of a data from being constructed into an index to being able to be searched is as:

t — t +T +T * insert _to _segment copy_segment ¹ reload _segment

T

Where, refers to time of constructing the data into the index segment, i.e., time of constructing the data into a primary index segment in this example of the présent disclosure, équivalent to a time-consuming of constructing one data into an index segment plus a time-consuming of merging the index segment into the primary index segment; in a system of ten T millions of standalones, the time-consuming is at a level of seconds. refers to a time-consuming of copying one primary index segment; in a system of ten millions of standalones T and 1000M local area network, the time-consuming is at a level of seconds, _ref_ers to a time-consuming of initializing a primary index segment, which can be almost 0s according to results of practical statistics. Thus, according to the above analysis, the total delay of the search method provided in this example is at a level of seconds; it can be seen, the search speed of the search method provided in this example is quite fast, and there is significant improvement compared with the related art.

In the method provided in this example, through obtaining index segments from the index constructing device and receiving primary index segments or the large index segments retumed and constructed by the index constructing device, the search service can be provided and search speed can be improved. When merging the first-level index segments into the corresponding second-level index segment, the first-level index segments are preferably merged into a second-level index segment neighboring the free memory area, thereby improving utilization ofthe memory. Further, the search method provided in this example also adopts a fault-tolerant mechanism for abnormal situations encountered in the searching processing, thereby ensuring real-time performance of searching.

Example six

This example provides a search device, refemng to Fig. 12, the device includes:

an obtaining module 1201 configured to, every each preset period, obtain a first-level index segment from an index constructing device, and merge the obtained first-level index segment into a first-level index segment set;

a second merging module 1202 configured to copy each first-level index segment in the first-level index segment set obtained by the obtaining module 1201, and merge the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

a search module 1203 configured to provide search service according to the first-level index segments in the first-level index segment set obtained by the obtaining module 1201 and the second-level index segments in the second-level index segment set obtained by the second merging module 1202.

Specifically, the second merging module 1202 is specifically configured to çopy each first-level index segment in the first-level index segment set in turn according to time sequence, and merge the copied first-level index segments into the corresponding second-level index segment. The corresponding second-level index segment is a second-level index segment which is in the second-level index segment set and neighbors a free memory.

Each of the first-level index segment obtained by the obtaining module 1201 and the second-level index segment obtained by the second merging module 1202 includes a sequence number value and a sign of whether being complété;

Accordingly, referring to Fig. 13, the obtaining module 1201 specifically indudes:

a second sending unit 1201a configured to, every each preset period, send an index segment application to the index constructing device, the application carrying a sequence number value and a sign of whether being complété;

a second receiving unit 1201b configured to receive the index segment sent by the index constructing device according to the sequence number value and sign of whether being complété sent by the second sending unit 1201a, and take the received index segment as a first-level index segment and merge the first-level index segment into a corresponding second-level index segment;

Accordingly, when the second merging module 1202 copies each first-level index segment in the first-level index segment set obtained by the obtaining module 1201, and merges the copied first-level index segments into the corresponding second-level index segment, a sequence number value of the second-level index segment after merged is the larger sequence number value of a sequence number value of the first-level index segment and a sequence number value of the second-level index segment when merging.

In the device provided in this example, through obtaining index segments from the index constructing device and receiving primary index segments or the large index segments retumed and constructed by the index constructing device, the search service can be provided and search speed can be improved. When merging the first-level index segments into the corresponding second-level index segment, the first-level index segments are preferably merged into a second-level index segment neighboring the free memory area, thereby improving utilization of the memory. Further, the device provided in this example also adopts a fault-tolerant mechanism for abnormal situations encountered in the searching processing, thereby ensuring real-time performance of searching.

Example seven

This example provides a search system, referring to Fig. 14, the system indudes an index constructing device 1401 and a search device 1402.

The index constructing device 1401 can be the index constructing device provided in the above example three.

The search device 1402 can be the search device provided in the above example six.

In actual applications, the search system indudes the index construding device 1401 and the search device 1402; the index constructing device 1401 constructs the obtained data into an index segment and sends the constructed index segment to the search device 1402, or the search device 1402 actively obtains index segments from the index constructing device 1401 and process the index segments, so as to provide search service for usera. Generally, one index constructing device 1401 can provide service for a plurality of search devices 1402. In large-scale search applications, data can be eut into multiple parts, and each part of data corresponding to one index constructing device 1401 and a plurality of the search devices 1402. This example and each of the above examples are illustrated for only one part of data; with respect to support large-scale search applications, it is only needed to copy multiple copies of the system provided in this example with each copy storing different data and merge these data again when searching. The index constructing device 1401 can be in a separate device and each device for searching only contains one search device 1402, information can be transmitted between the device containing index constructing device 1401 and the device containing search device 1402 through local area network. Thus, in the search system provided in this example, there can be one or more the search devices 1402, then the structure of the search system can also be as shown in Fig. 15.

The system provided in this example uses the index constructing device to construct a plurality of primary index segments on the basis of time, merge the constructed primary index segments into the large index segment and send the primary index segments or the large index segments to the search device so as to provide search service, thus, search speed can be improved. When the index constructing device merges the index segments, the index segments are preferably merged into an index segment neighboring the free memory area, thereby improving utilization of the memory.

It should be noted, when the index constructing device provided in the above example constructs indexes and when the search device provided in the above example performs searching, the above division of functional modules is only taken as an example for illustration; in actual applications, the above functions can be allocated to be completed by different functional modules according to needs, i.e., dividing the internai structure of the device into different functional modules so as to complété ail or part of the above-described functions. Further, the index constructing device provided in the above example and the example of the index constructing method belong to the same concept, and the search device provided in the above example and the example of the search method belong to the same concept, the spécifie implémentation processes refer to the examples of methods and are not repeated here.

The above serial numbera of the examples of the présent disclosure are just used for description, rather than indicating the priorities thereof.

Ail or parts of steps of the examples of the présent disclosure may be implemented by hardware, or may be completed by using a program to instruct relevant hardware. The program may be stored in readable storage medium, such as CD-ROM or hard disk, etc.

The foregoing are only preferred examples of the présent disclosure, and are not used to limit the présent disclosure. Any modification, équivalent replacement, or improvement made without departing from the spirit and principle of the présent disclosure should fall within the scope of the 5 présent disclosure.

Claims (15)

1. 5 What is Claimed is:

   1. An index constructing method comprising:

   constructing data obtained in each preset time period into a primary index segment, respectively, and merging the constructed primary index segment into a primary index segment set after each preset time period is over;

   10 copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, to obtain a large index segment set; and sending the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device to provide search 15 service;

   wherein the copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment specifically includes:

   copyingeach primary index segment in the primary index segment setin tum 20 according to time sequence, and merging the copied primary index segments into the corresponding large index segment; wherein the corresponding large index segment is a large index segment which is in the large index segment set and neighbors a free memory.
2. 2. The method of claim 1, wherein after merging the constructed primary index segments into a corresponding large index segment, the method further includes:

   25 every each preset period, deleting primary index segments which are constructed before the each preset period from the primary index segment set.
3. 3. The method of claim 1 or 2, wherein after the copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, to obtain a large index segment set,

   30 the method further includes:

   backing up each large index segment in the large index segment set.
4. 4. The method of claim 3, wherein each of the primary index segment and the large index segment includes a sequence number value and a sign of whether being complété;

   when copying each primary index segment in the primary index segment set and merging the copied primary index segments into a corresponding large index segment, asequence number value of the large index segment after merged is the larger sequence number value of a sequence number value of the primary index segment and a sequence number value of the large index segment when merging;

   the sending the primary index segments in the primary index segment set or the large index segments in the large index segment set to a search device to provide search service specifically includes:

   receiving an index segment application sent from the search device, the application carrying a sequence number value and a sign of whether being complété;

   sending corresponding index segments to the search device according to the sequence number value and sign of whether being complété carried in the application, the corresponding index segments being primary index segments of the primary index segment set or large index segments of the large index segment set.
5. 5. The method of claim 4, wherein the retuming corresponding index segments to the search device according to the sequence number value and sign of whether being complété carried in the application, specifically includes:

   if the sequence number value carried in the application is null, reading backed up index segmentsand sending the backed up index segments to the search device;

   if the sequence number value carried in the application is non-null and the application carries a sign of being incomplète, retuming an index segment with a sequence number value the same as the sequence number value carried in the application to the search device;

   if the sequence number value carried in the application is non-null and the application carries a sign of being complété, retuming an index segment with a sequence number value which is the next sequence number value of the sequence number value carried in the application to the search device.
6. 6. An index constructing device comprising:

   a constructing module configured to construct data obtained in each preset time period into a primary index segment, respectively, and merge the constructed primary index segment into a primary index segment set after each preset time period is over;

   a first merging module configured to copy each primary index segment in the primary index segment set obtained by the constructing moduleand merge the copied primary index segments into a corresponding large index segment to obtain a large index segment set; and a sending module configured to send the primary index segments in the primary index segment set obtained by the constructing module or the large index segments in the large index segment set obtained by the first merging module to a search device to provide search service;

   wherein the first merging module is specifically configured to copy, in tum according to time sequence, each primary index segment in the primary index segment set obtained by the constructing module through merging, and merge the copied primary index segments into a corresponding large index segment which is a large index segment in the large index segment set and neighbors a free memory.
7. 7. The device of claim 6, wherein the device further includes:

   a délétion module configured to, after the constructing module merges the constructed primary index segment into the primary index segment set, every each preset period, delete primary index segments which are constructed before the each preset period from the primary index segment set.
8. 8. The device of claim 6 or 7, wherein the device further includes:

   a backup module configured to, after the first merging module copies each primary index segment in the primary index segment set obtained by the constructing moduleand merges the copiedprimary index segments into a corresponding large index segment to obtain a large index segment set, back up each large index segment in the large index segment set.
9. 9. The device of claim 8, wherein each of the primary index segment constructed by the constructing module and the large index segment obtained by the first merging module through merging includes a sequence number value and a sign of whether being complété;

   when the first merging module copies each primary index segment in the primary index segment set obtained by the constructing moduleand merges the copiedprimary index segments into a corresponding large index segment, asequence number value of the large index segment after merged is the larger of a sequence number value of the primary index segment and a sequence number value of the large index segment when merging;

   thesending module specifically includes:

   a first receiving unit configured to receive an index segment application sent from the search device, the application carrying a sequence number value and a sign of whether being complété;

   a first sending unit configured to, send corresponding index segments to the search device according to the sequence number value and sign of whether being complété carried in the application received by the first receiving unit, the corresponding index segments being primary index segments of the primary index segment set or large index segments of the large index segment set.
10. 10. The device of claim 9, whereinthe first sending unit is specifically configured to, if the sequence number value carried in the application received by the first receiving unit is null, read backed up index segments and send the backed up index segments to the search device; if the sequence number value carried in the application received by the first receiving unit is non-null and the application carries a sign of being incomplète, retum an index segment with a sequence number value the same as the sequence number value carried in the application to the search device; if the sequence number value carried in the applicationreceived by the first receiving unit is non-null and the application carries a sign of being complété, retum an index segment with a sequence number value which is the next sequence number value of the sequence number value carried in the application to the search device.
11. 11. A search method comprising:

    every each preset period, obtaining a first-level index segment from an index constructing device, and merging the obtained first-level index segment into a first-level index segment set;

    copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

    providing search service according to the first-level index segments in the first-level index segment set and the second-level index segments in the second-level index segment set;

    whereineach of the first-level index segment and the second-level index segment includes a sequence number value and a sign of whether being complété;

    every each preset period, obtaining a first-level index segment from an index constructing device, specifîcally includes:

    every each preset period, sendîng an index segment application to the index constructing device, the application carrying a sequence number value and a sign of whether being complété; the sequence number value being the largest sequence number value of those of already existed index segments or null;

    receiving the index segment sent by the index constructing device according to the sequence number value and sign of whether being complété, and taking the received index segment as a first-level index segment;

    when copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment, asequence number value of the second-level index segment after merged is the largersequence number value of a sequence number value of the first-level index segment and a sequence number value of the second-level index segment when merging.
12. 12. The method of claim 11, wherein the copying each first-level index segment in the first-level index segment set and merging the copied first-level index segments into a corresponding second-level index segment, specifîcally includes:

    copying each first-level index segment in the first-level index segment set in tum according to time sequence and merging the copied first-level index segments into a corresponding second-level index segment; wherein the corresponding second-level index segment is a second-level index segment which is in the second-level index segment set and neighbors a free memory.
13. 13. A search device comprising:

    an obtaining module configured to, every each preset period, obtain a first-level index segment from an index constructing device, and merge the obtained first-level index segment into a first-level index segment set;

    a second merging module configured to copy each first-level index segment in the first-level index segment set obtained by the obtaining module, and meige the copied first-level index segments into a corresponding second-level index segment to obtain a second-level index segment set;

    a search module configured to provide search service according to the first-level index segments in the first-level index segment set obtained by the obtaining module and the second-level index segments in the second-level index segment set obtained by the second merging module;

    whereinthe second merging module is specifically configured to copy each first-level index segment in the first-level index segment set in tum according to time sequence, and merge the copied first-level index segments into a corresponding second-level index segment; the corresponding second-level index segment is a second-level index segment which is in the second-level index segment set and neighbors a free memory.
14. 14. The device of claim 13, wherein each of the first-level index segment obtained by the obtaining module and the second-level index segment obtained by the second merging module includes a sequence number value and a sign of whether being complété;

    the obtaining module specifically includes:

    a second sending unit configured to, every each preset period, send an index segment application to the index constructing device, the application carrying a sequence number value and a sign of whether being complété;

    a second receiving unit configured to receive the index segment sent by the index constructing device according to the sequence number value and sign of whether being complété sent by the second sending unit, and take the received index segment as a first-level index segment and merge the first-level index segment into a corresponding second-level index segment;

    when the second merging module copies each first-level index segment in the first-level index segment set obtained by the obtaining module and merges the copied first-level index segments into the corresponding second-level index segment, asequence

    5 number value of the second-level index segment after merged is the largersequence number value of a sequence number value of the first-level index segment and a sequence number value of the second-level index segment when merging.
15. 15. A search system comprising an index constructing device and more than one search devices; wherein

    10 the index constructing device is anindex constructing device of any one of claim from 6 to 10;

    the search device is a search device of claim 13 or 14.