Method and/or system for performing tree matching

Abstract

Embodiments of methods, apparatuses, devices and/or systems for performing tree matching are disclosed.

Description

BACKGROUND

This disclosure is related to tree matching.

In a variety of fields, data or a set of data, may be represented in a hierarchical fashion. This form of representation may, for example, convey information, such as particular relationships or patterns between particular pieces of data or groups of data and the like. However, manipulating and/or even recognizing specific data representations or patterns is not straight-forward, particularly where the data is arranged in a complex hierarchy. Without loss of generality, examples may include a database, and further, without limitation, a relational database. Techniques for performing operations on such databases or recognizing specific patterns, for example, are computationally complex, time consuming, and/or otherwise cumbersome. A need, therefore, continues to exist for improved techniques for performing such operations and/or recognizing such patterns.

BRIEF DESCRIPTION OF THE DRAWINGS

Subject matter is particularly pointed out and distinctly claimed in the concluding portion of the specification. The claimed subject matter, however, both as to organization and method of operation, together with objects, features, and advantages thereof, may best be understood by reference of the following detailed description when read with the accompanying drawings in which:

FIG. 1 is a schematic diagram of one embodiment of a tree;

FIG. 2 is a schematic diagram illustrating one embodiment of an ordered binary edge labeled tree;

FIG. 3 is a schematic diagram illustrating another embodiment of an ordered binary edge labeled tree;

FIG. 4 is a schematic diagram illustrating an embodiment of a binary edge labeled string;

FIG. 5 is a table illustrating an embodiment of an association between natural numerals and unordered BELTs;

FIG. 6 is a schematic diagram of an embodiment of a binary node labeled tree;

FIG. 7 is a schematic diagram illustrating another embodiment of a binary node labeled tree;

FIG. 8 is a schematic diagram illustrating an embodiment of an inversion operation and an embodiment of a merger operation applied to an embodiment of ordered binary edge labeled trees;

FIG. 9 is a schematic diagram illustrating examples of potential embodiments of query and target trees;

FIGS. 10a and 10b are schematic diagrams illustrating, respectively additional examples of potential embodiments of query and target trees;

FIG. 11 is a schematic diagram illustrating an embodiment of a match and a non-match for an embodiment of an ordered binary edge labeled tree; and

FIG. 12 is a schematic diagram illustrating, for the target tree example of FIG. 10b, potential subtree examples to be employed to accomplish matching for an embodiment of a method of performing tree matching.

DETAILED DESCRIPTION

In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the claimed subject matter. However, it will be understood by those skilled in the art that the claimed subject matter may be practiced without these specific details. In other instances, well-known methods, procedures, components and/or circuits have not been described in detail so as not to obscure the claimed subject matter.

Some portions of the detailed description which follow are presented in terms of algorithms and/or symbolic representations of operations on data bits or binary digital signals stored within a computing system, such as within a computer or computing system memory. These algorithmic descriptions and/or representations are the techniques used by those of ordinary skill in the data processing arts to convey the substance of their work to others skilled in the art. An algorithm is here, and generally, considered to be a self-consistent sequence of operations and/or similar processing leading to a desired result. The operations and/or processing involve physical manipulations of physical quantities. Typically, although not necessarily, these quantities may take the form of electrical and/or magnetic signals capable of being stored, transferred, combined, compared and/or otherwise manipulated. It has proven convenient, at times, principally for reasons of common usage, to refer to these signals as bits, data, values, elements, symbols, characters, terms, numbers, numerals and/or the like. It should be understood, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels. Unless specifically stated otherwise, as apparent from the following discussion, it is appreciated that throughout this specification discussions utilizing terms such as “processing”, “computing”, “calculating”, “determining” and/or the like refer to the actions and/or processes of a computing platform, such as a computer or a similar electronic computing device, that manipulates and/or transforms data represented as physical electronic and/or magnetic quantities and/or other physical quantities within the computing platform's processors, memories, registers, and/or other information storage, transmission, and/or display devices.

In a variety of fields, data or a set of data, may be represented in a hierarchical fashion. This form of representation may, for example, convey information, such as particular relationships or patterns between particular pieces of data or groups of data and the like. However, manipulating and/or even recognizing specific data representations or patterns is not straight-forward, particularly where the data is arranged in a complex hierarchy. Without loss of generality, examples may include a database and further, without limitation, a relational database. Techniques for performing operations on such databases or recognizing specific patterns, for example, are computationally complex, time consuming, and/or otherwise cumbersome. A need, therefore, continues to exist for improved techniques for performing such operations and/or recognizing such patterns.

As previously discussed, in a variety of fields, it is convenient and/or desirable to represent data, a set of data and/or other information in a hierarchical fashion. In this context, such a hierarchy of data shall be referred to as a “tree.” In a particular embodiment, a tree may comprise a finite, rooted, connected, acyclic graph. Likewise, such trees may be either ordered or unordered. Here, ordered refers to the notion that there is an ordering or precedence among nodes attached to a common node corresponding to the order of the attached nodes shown in a graphical illustration. An ordered tree is illustrated here, for example, in FIG. 1 by embodiment 100. As illustrated, the root of this particular embodiment encompasses node 105. In addition to 105, there are eight other nodes designated 110 to 145, respectively. Likewise, the nodes are connected by branches referred to, in this context, as edges. Thus, the nodes of this tree are connected by eight edges. This embodiment, therefore, illustrates a finite tree that is rooted by node 105. Furthermore, the nodes are connected, meaning, in this context, that a path exists between any two nodes of the tree. The tree is likewise acyclic, meaning here, that no path in the tree forms a complete loop.

As previously suggested, in a variety of contexts, it may be convenient and/or desirable to represent a hierarchy of data and/or other information using a structure, such as the embodiment illustrated in FIG. 1. One particular embodiment, without loss of generality, of a tree may include edges that are labeled with data and/or other values. Likewise, in one particular embodiment, such data and/or values may be limited to binary data, that is, in this example, either a binary one or a binary zero. Here, such an embodiment may be referred to as a binary edge labeled tree (BELT), as shall be discussed in more detail hereinafter. In this embodiment, an ordered binary edge labeled tree is shown. Here, ordered refers to the notion that there is an ordering or precedence among nodes attached to a common node corresponding to the order of the attached nodes shown in a graphical illustration.

One example of an ordered BELT is illustrated by embodiment 200 of FIG. 2. Thus, as illustrated, the edges of the BELT shown in FIG. 2, including nodes 205, 210, 215, 220, 225, 230, 235, 240 and 245, are labeled with either a binary zero or binary one. FIG. 3 illustrates another embodiment 300 of a different ordered binary edge labeled tree, including nodes 305, 310, 315, 320, 325, 330, 335, 340 and 345. It is noted that this tree is similar or isomorphic in arrangement or structure to the embodiment of FIG. 2, as shall be explained in more detail hereinafter.

A subset of BELTs may be referred to, in this context, as binary edge labeled strings (BELSs). One embodiment, 400, is illustrated in FIG. 4. Thus, as illustrated by embodiment 400, this particular binary edge labeled string comprises four nodes and three edges, where the edges are labeled, respectively, binary zero, binary one and binary zero. Thus, a binary edge labeled string comprises a binary edge labeled tree in which each node has no more than two edges. To reiterate, in this context, a string comprises a binary edge labeled string and a tree comprises a binary edge labeled tree if each edge of the string or, tree respectively stores a single bit. Likewise, in this context, two nodes are employed to support an edge holding a single piece of binary data. At this point, it is worth noting that strings and trees having nodes and edges, such as previously described, may be represented in a computing platform or similar computing device through a data structure or a similar mechanism intended to capture the hierarchical relationship of the data, for example. It is intended that all such embodiments are included within the scope of the claimed subject matter.

As may be apparent by a comparison of FIG. 4 with, for example, FIG. 2 or FIG. 3, typically, a binary edge labeled tree has the ability to be richer and convey more data and/or more information than a binary edge labeled string. This may be observed, by a comparison of FIG. 4 with, for example, FIG. 2 or FIG. 3. Of course, depending on the particular tree and the particular string, there may be contrary examples, such as where the string is particularly large and the tree is particularly small. The aspect of BELTs to be richer in information may be one potential motivation to employ BELTs over BELSs, for example.

Despite the prior observation, as shall be described in more detail hereinafter, an association may be made between any particular binary edge labeled string and a binary edge labeled tree or vice-versa, that is, between any particular binary edge labeled tree and a binary edge labeled string. See, for example, U.S. provisional patent application Ser. No. 60/543,371, filed on Feb. 9, 2004, titled “Manipulating Sets of Hierarchical Data,” assigned to the assignee of the presently claimed subject matter. In particular, an association may be constructed between binary edge labeled trees and binary edge labeled strings by enumerating in a consecutive order binary edge labeled strings and binary edge labeled trees, respectively, and associating the respectively enumerated strings and trees with natural numerals. Of course, many embodiments of associations between trees, whether or not BELTS, and strings, whether or not BELS, or between trees, whether or not BELTs, and natural numerals are possible. It is intended that the claimed subject matter include such embodiments, although the claimed subject matter is not limited in scope to the aforementioned provisional patent application or to employing any of the techniques described in the aforementioned provisional patent application.

Binary edge labeled trees may also be listed or enumerated. See, for example, previously cited U.S. provisional patent application Ser. No. 60/543,371. This is illustrated, here, for example, in FIG. 5. It is noted that this particular figure also includes the associated natural numerals. The association 500 of such numerals for this particular embodiment should be clear based at least in part on previously cited U.S. provisional patent application Ser. No. 60/543,371. However, it is, of course, again noted that the claimed subject matter is not limited in scope to employing the approach or approaches described in U.S. provisional patent application Ser. No. 60/543,371. U.S. provisional patent application Ser. No. 60/543,371 is provided simply as an example of listing or enumerating unordered BELTs. Thus, it is noted further that the BELTs described are unordered.

However, for this particular embodiment, although the claimed subject matter is not limited in scope in this respect, a method of enumerating a set of ordered trees may begin with enumeration of an empty binary edge labeled tree and a one node binary edge labeled tree. Thus, the empty tree is associated with the natural numeral zero and has a symbolic representation as illustrated in FIG. 5 (circle). Likewise, the one node tree, which holds no data, is associated with the natural numeral one and has a graphical representation of a single node. For higher positive natural numbers, ordered trees may be generated by a process described, for example, in “The Lexicographic Generation of Ordered Trees,” by S. Zaks, The Journal of Theoretical Computer Science, Vol. 10(1), pp 63-82, 1980, or, “Enumerating Ordered Trees Lexicographically,” by M. C. Er, Computation Journal, Vol. 28, Issue 5, pp 538-542, 1985. This may be illustrated, for example in FIG. 5, as described in more detail below.

As illustrated, for this particular embodiment, and as previously described, the empty tree has zero nodes and is associated with the natural numeral zero. Likewise, the one node tree root comprises a single node and is associated with the natural numeral one. Thus, to obtain the tree at position two, a root node is attached and connected to the prior root node by an edge. Likewise, here, by convention, the edge is labeled with a binary zero. If, however, the tree formed by the immediately proceeding approach were present in the prior enumeration of trees, then a similar process embodiment is followed, but, instead, the new edge is labeled with a binary one rather than a binary zero. Thus, for example, to obtain the binary edge labeled tree for position three, a new root node is connected to the root node by an edge and that edge is labeled with a binary one.

Continuing with this example, to obtain the binary edge labeled tree for position four, observe that numeral four is the product of numeral two times numeral two. Thus, a union is formed at the root of two trees, where, here, each of those trees is associated with the positive natural numeral two. Likewise, to obtain the binary edge labeled tree for position five, begin with the binary edge labeled tree for position two and follow the previously articulated approach of adding a root and an edge and labeling it with a binary zero.

In this context, adding a root node and an edge and labeling it binary zero is referred to as a “zero-push” operation and adding a root node and an edge and labeling it binary one is referred to as a “one-push” operation. Thus, referring again to FIG. 5, the one-push of the root tree is the tree at position three. This follows from FIG. 9 of previously referenced U.S. provisional patent application Ser. No. 60/543,371, since P(2*1)=P(2)=3. Likewise, the tree at position five is the zero-push of the tree at position 2. Again, this follows from FIG. 9 of the previously referenced US provisional patent application, since P(2*2−1)=P(3)=5.

In the embodiment just described, binary edge labeled trees use binary numerals “0” and “1.” However, the claimed subject matter is not limited in scope to binary edge labeled trees. For example, trees may employ any number of numeral combinations as labels, such as triplets, quadruplets, etc. Thus, using a quadruplet example, it is possible to construct trees, such as a zero-push of a particular tree, a one-push of that tree, a two-push of that tree, and a three-push of that tree. Thus, for such trees, edges may be labeled 0, 1, 2 or 3, etc.

The foregoing discussion has begun to characterize an algebra involving trees, in this particular embodiment, an algebra for ordered binary edge labeled trees or ordered BELTs. The foregoing discussion, therefore, defines a value zero, a zero node tree for this particular embodiment, value one, a one node tree for this particular embodiment, and a monadic operation, previously described as zero-push. For this particular embodiment, the push operation shall also be referred to as the successor operation. For this particular embodiment, this shall be denoted as S(x), where x refers to the tree to which the successor operation is applied. Of course, the claimed subject matter is not limited in scope to the 1-5 successor operation, S(x), being limited to a zero-push. For example, alternatively, a “one-push” may be employed. For this embodiment, this is analogous, for example, to the convention that “0” represent “off” and “1” represent “on.” Alternatively and equivalently, “1” may be employed to represent “off,” and “0” may be employed to represent “on,” without loss of generality.

For this particular embodiment, two additional operations may be characterized, an “inversion” operation and a “merger” operation. For this particular embodiment, the inversion operation, when applied to a binary edge labeled tree, such as an ordered BELT, refers to replacing a “1” with a “0” and replacing a “0” with a “1”. Likewise, the merger operation with respect to trees refers to merging two trees at their roots. These two operations are illustrated, for example, at embodiment 800 in FIG. 8, showing a probe tree P in embodiment 805 and a target tree T in embodiment 810.

As will now be appreciated, the inversion operation comprises a monadic operator while the merger operation comprises a binary operator. Likewise, the constants zero/one, referred to above, may be viewed as an operation having no argument or as a zero argument operator or operation. Thus, this operation, in effect, returns the same value whenever applied. Here, for this particular embodiment, the constant value zero, or zero argument operation that returns “0,” is denoted as “c,” the merger operator is denoted as “*”, the inversion operation is denoted as “′”, and the successor operator is denoted as previously described.

One additional aspect of the foregoing relationships that was omitted from this embodiment, but that might be included in alternate embodiments, is the addition of a second monadic operator, denoted here as “T(x).” This particular operator is omitted here without loss of generality at least in part because it may be defined in terms of operators previously described. More particularly, T(x)=S(x′)′, may be included in alternate embodiments. This approach, though not necessary from an implementation perspective, may add some symmetry and elegance to the above basis relationships. For example, it may be demonstrated that S(x)′=T(x′) and S(x′)=T(x)′. In some respects, this relationship is analogous to the relationship between the logical operations OR and AND in Boolean algebra, where −(A AND B)=−A OR −B, and −(A OR B)=−A AND −B. However, as indicated above, this may be omitted without loss of generality and, therefore, for implementation purposes, it may be easier to implement four operators rather than five.

Of course, as previously alluded to, for this particular embodiment, a useful distinction is also made between an ordered binary edge labeled tree and an unordered binary edge labeled tree. In this context, and as previously suggested, the notion of “ordered” refers to the property that the nodes attached to a particular node form an ordered set, the order corresponding to the order in which those nodes are displayed in the graph of the tree. However, it may likewise be observed that two ordered trees are resident in the same equivalence class of unordered BELTs if and only if the two trees are commutative translates of each other. In other the words, the two trees are equivalent and in the same unordered BELT equivalence class where the trees differ only in the order of the attached nodes.

Although the claimed subject matter is not limited in scope in this respect, one technique for implementing this approach may be to apply a table look up approach. Techniques for performing table look-ups are well-known and well-understood. Thus, this will not be discussed in detail here. However, it shall be appreciated that any and all of the previously described and/or later described processing, operations, conversions, transformations, manipulations, etc. of strings, trees, numerals, data, etc. may be performed on one or more computing platforms or similar computing devices, such as those that may include a memory to store a table as just described, although, the claimed subject matter is not necessarily limited in scope to this particular approach. Thus, for example, a hierarchy of data, such as a tree as previously described, for example, may be formed. Likewise, operations and/or manipulations, as described, may be performed; however, operations and/or manipulations in addition to those described or instead of those described may also be applied. It is intended that the claimed subject matter cover such embodiments.

Embodiments of a method of performing tree matching has a variety of potentially useful applications. As described previously, trees provide a technique for structuring and/or depicting hierarchical data. Thus, for example, trees may be employed to represent language sentence structures, computer programs, algebraic formulae, molecular structures, family relationships and more. For example, one potential application of such a tree reduction technique is in the area of pattern matching. Thus, in pattern matching, substructures, in the form of a tree, for example, may be located within a larger structure, also in the form of a tree, referred to in this context as the target. This may be accomplished by comparing the structures; however, typically, such a comparison is complex, cumbersome, and/or time consuming. Although the claimed subject matter is not limited in scope to pattern matching or to any of the other potential applications described above, it may be instructive to work through at least one particular example of applying the previously described tree reduction approach to a pattern matching problem to demonstrate the power and/or versatility of this particular embodiment.

Within this particular context and for this particular embodiment, there are a number of potential pattern matching inquiries that may be made. Although these are simply examples and the claimed subject matter is not limited in scope to only these particular inquiries, one such inquiry, for example, may be whether a first tree, such as an ordered binary edge labeled tree, is equal to a second binary edge labeled tree? To phrase this differently, it may be useful to determine whether the trees match exactly. Likewise, another such query, or active verb, may be referred to in this context as a rooted partial sub tree (RPS) query or inquiry. This particular type of query or inquiry is demonstrated with reference to embodiment 1100FIG. 11 in examples 1205 and 1210.

Thus, in Examples 1 and 2 of FIG. 11, the right-hand sides depict a binary edge labeled tree for the numeral 60543371. See, for example, the previously referenced U.S. provisional patent application 60/543,371. Here, in Example 1, the left-hand side of FIG. 11 provides a rooted partial subtree of the right-hand side.

In this context, the term rooted refers to a comparison in which the roots of the left-hand side and the right-hand side are matched or compared. The notion of a partial subtree is to be distinguished from the notion of a full subtree. In this context, therefore, a rooted full subtree refers to the equality described above. Likewise, then, a rooted partial subtree refers to a match with another tree, but only to the extent of the nodes and edges present for the rooted partial subtree. Thus, the target may contain additional nodes, edges, and/or labels that are omitted from the rooted partial subtree. By way of contrast, Example 2 demonstrates on the left-hand side a tree that is not a rooted partial subtree of the right-hand side tree, although the left-hand side tree has the same arrangement of nodes and edges as a rooted partial subtree of the right-hand side. Thus, another type of match may occur where the arrangement of the nodes and edges match, but the labels do not match, as in Example 2.

One query or question to be posed, for the purposes of pattern matching, is whether the tree on the left-hand side, such as in example one, is a rooted partial subtree of the tree on the right-hand side. In addition to that, several other potential questions may be posed and potentially answered. For example, if the tree on the left-hand side is a rooted partial subtree of the tree on the right-hand side, it may be useful to know how many times this rooted partial subtree is present in the right-hand side tree. Likewise, assume that a rooted partial subtree is present more than once. It may be useful to have a mechanism to identify one of the several rooted partial subtrees to a machine, for example, for further processing.

It also may be desirable, in other circumstances, to determine whether there is a match between a rooted tree and a subtree that is not rooted. In this context this may be referred to, for example, as a “projected match”. In this context, this refers to projecting one tree into another tree without matching corresponding roots and having the form and labels of the projected tree still be preserved in the tree in which it is projected.

Likewise, with reference to Example 2, in which the tree on the left-hand side does not match the tree on the right-hand side, an alternative query or question may relate to a measurement of the similarities and/or differences, as an embodiment of a measurement of the matching. For example, particular branches of the tree on the left-hand side may match with particular branches of the tree on the right-hand side, although overall, the entire tree on the left-hand side may not match to a subportion of the tree on the right-hand side, in this particular example. Thus, it may be appropriate, for example, to weight the matching in some form. Such an approach, for example, might be employed in data analysis, as simply one example. In one embodiment, for example, it may be desirable to identify a partial match that results in the maximum number of matching nodes and edges; likewise, in a different embodiment, it may be desirable to identify a partial match such that the match is closest to or most remote from the root. Again, any one of a number of other approaches is possible and such approaches included within the scope of the claimed subject matter. Thus, it may be desirable, assuming there is no identical match, to identify the closest match where “closest” or “most remote” is defined with respect to a particular weighted criterion designed to achieve a particular objective, such as the examples previously described.

Furthermore, to apply such queries such as, for example, determining whether a first tree is a rooted partial subtree of another tree, as indicated by the tree expression above, involves the application of known programming techniques. See, for example, Chapter 4, “Tree Isomorphism,” of Algorithms on Trees and Graphs, by Gabriel Valiente, published by Springer, 2002. Such well-known and well-understood programming techniques will not be discussed here in any detail.

Much of the prior discussion was provided in the context of ordered binary edge labeled trees. However, a similar approach may be applied to unordered binary edge labeled trees, for example. In general, it is understood that performing such simplifications or reductions to unordered BELTs presents more of a processing challenge. See, for example, “Tree Matching Problems with Applications to Structured Text Databases,” by Pekka Kilpelainen, Ph.D dissertation, Department of Computer Science, University of Helsinki, Finland, November, 1992. A potential reason may be that a greater number of possibilities are present combinatorially in those situations in which nodes may be unordered rather than ordered.

Of course, the claimed subject matter is not limited to ordered or unordered binary edge labeled trees. For example, as described in previously cited U.S. provisional patent application No. 60/543,371, binary edge labeled trees and binary node labeled trees may be employed nearly interchangeably to represent substantially the same hierarchy of data. In particular, a binary node labeled tree may be associated with a binary edge labeled tree where the nodes of the binary node labeled tree take the same values as the edges of the binary edge labeled tree, except that the root node of the binary node labeled tree may comprise a node having a zero-value or a null value. This is illustrated, for example, in embodiment 600 of FIG. 6. Thus, rather than employing binary edge labeled trees, the previously described embodiments may alternatively be performed using binary node labeled trees. As one example embodiment, operations and/or manipulations may be employed using binary edge labeled trees and the resulting binary edge labeled tree may be converted to a binary node labeled tree. However, in another embodiment, operations and/or manipulations may be performed directly using binary node labeled trees where a different association embodiment is employed.

As previously alluded to, tree matching has a variety of potential and useful applications. It is noted that the claimed subject matter is not limited in scope to any particular set of applications. It is intended that the claimed subject matter include all currently known applications and all future developed applications. However, one aspect of tree matching relates to recognizing specific patterns. In particular, as previously alluded to, it may be desirable, depending at least in part on the circumstances of problem, to match a partial subtree, referred to here a query tree, with another tree, referred to here as a target tree. For this particular example or embodiment, the trees comprise unordered binary edge labeled trees, although, of course, the claimed subject matter is not limited in scope to these particular types of trees.

As an example, it may be desirable for a query tree, designated P, for example, to be matched against a target tree, designated T, for example. Thus, as previously suggested, it may be desirable to identify within T the correspondence of the node/label set of P within the node/label set of T that preserves the structure and labels of P. Likewise, as previously indicated, it is also possible that multiple occurrences of P may be found in T. Thus, the task, for this particular embodiment, is to identify and count the number of matches of P into T. Embodiment 900 of FIG. 9, for example, illustrates a query tree P with two partial subtree matches 910 and 915 in a target tree T from among query trees 905.

Without belaboring the discussion, as discussed in, U.S. provisional patent application No. 60/575,784, filed on May 28, 2004, by J. J. LeToumeau, titled, METHOD AND/OR SYSTEM FOR SIMPLIFYING TREE EXPRESSIONS, SUCH AS FOR PATTERN MATCHING, assigned to the assignee of the presently claimed subject matter, techniques may be applied to perform such tree matching operations. However, as the complexity of these trees increase, such as, for example, the tree depth and number of nodes, natural numerals that may be employed as part of the tree matching operations, such as those, for example, corresponding to the trees themselves, for example, U.S. provisional patent application No. 60/543,371, filed on Feb. 9, 2004, by J. J. LeToumeau, titled, MANIPULATING SETS OF HIERARCHICAL DATA, assigned to the assignee of the presently claimed subject matter, may become large. For example, experimental investigations have suggested that the number of bits in the natural numeral corresponding to a particular tree may approximately equal the number of nodes in the tree.

As is well-known, computing platforms, such as computers or other computing devices, typically represent natural numerals internally as a platform native integer of fixed size in a binary format, most commonly either 32 or 64 bits, although, of course, the claimed subject matter is not limited in scope in this respect. In this particular context, the term platform native integer of fixed size refers to the size of the data registers for the particular computing device. To take advantage of the computational power engineered to be delivered by such platforms, therefore, it may be desirable to perform operations in a manner so that numerals employed to perform the operations do not exceed a native platform sizes. Therefore, a mechanism to enable manipulation of trees whose corresponding natural numerals exceed a platform native size while employing numerals that do not exceed such sizes is desirable.

One potential approach might include developing a multi-precision software solution to store and manipulate (e.g., perform basic arithmetic operations, such as add and divide) for natural numerals larger than the native integer size. However, such an approach has a disadvantage in that the overhead associated with representing and manipulating large integers in software may be significant. Likewise, such multi-precision arithmetic operations, such as the multiply operation, for example, may potentially run orders of magnitude slower than the corresponding hardware operations specifically designed for platform native sized integers, for example.

Another potential approach may include subdividing the query and target trees. In such an approach they may be input to the tree matching problem so as to render the query and target trees expressible as numerals within the parameters of the particular platform. The results of the tree matching problem for such pieces may likewise be combined so that the correct number of matches for the particular query and target trees is obtained. The advantage of this approach is that the tree matching mechanism need not be modified beyond changing the type of natural numerals associated with trees. Furthermore, it has the potential advantage that the cost of operations associated with splitting up trees so that platform native arithmetic operations are performed may prove more beneficial than the costs associated with a multi-precision software approach.

In one particular embodiment, although the claimed subject matter is not limited in scope in this respect, it may be desirable to satisfy the following conditions:

• • 1. Trees formed that are direct children of the root node of the target tree comprise a platform native integer numeral (hereinafter, referred to as a “Treenum”).
  • 2. Query trees comprise a platform native integer numeral.It is desirable that these conditions be met so that, for this embodiment, the design of the processor to specifically handle platform native integer numerals is exploited at least in part. For this embodiment, for a given edge attached to root, the set of all nodes and edges below that edge is referred to as “child of root.” A subtree of root comprises one or more children of root that, for this embodiment, satisfy condition 1 above.

The target tree T, in this embodiment, is subdivided into a set of subtrees of root. However, for this embodiment, children of root are merged to form the subtrees so that the resulting tree, while being below a specific threshold, such as, for example, a threshold number of nodes, nonetheless, preferably is not subdivided significantly beyond that sufficient to place it below the threshold. For example, target tree, T, may be checked to determine the number of direct children of root that may be combined together while still complying with condition 1 above, for this embodiment. Experimental results indicated that the number of nodes in a BELT is at least roughly correlated to the number bits in the corresponding Treenum, for example. FIGS. 10a and 10b provide examples of query and target trees to be used to demonstrate applying this particular embodiment 1000.

In embodiment 1200 shown in FIG. 12, subtree 1220, also designated T_X, comprises two children of root while 1230 and 1240, respectively designated T_Y and T_Z, comprise one child of root. The entire target tree, T, for this example, therefore, comprises the merger of the subtrees, as indicated by the following relationship:T=T_X*T_Y*T_Z  [1]

Partial subtrees of the target subtrees may be enumerated in this particular embodiment. Several references outline methods to enumerate partial subtrees. See, for example, Chapter 4, “Tree Isomorphism,” of Algorithms on Trees and Graphs, by Gabriel Valiente, published by Springer, 2002. Such well-known and well-understood programming techniques will not be discussed here in any detail.

For this particular embodiment, we define T_J to be any single element in the set of all target subtrees (e.g., in this example, T_X, T_Y, T_Z). Each T_J, for this embodiment, will then have as its root, the root node of T, called N_ROOT here. We further define a rooted partial subtree of any T_J as a partial subtree which has as its root N_ROOT. Likewise, we define non-rooted partial subtrees as all other partial subtrees. Thus, the enumerated partial subtrees for each T_J are divided into two categories: rooted and non rooted.

A distinction is made for this embodiment between rooted and non rooted partial subtrees at least in part because the non rooted matching case can be treated as a simple match of the whole query P against the list of non rooted subtrees for each T_J in succession. The rooted matching problem is more complex at least in part because multiple children of the query root node may match in multiple different T_J. To address this, an enumeration is made of combinations of query children matches into the list of rooted partial subtrees for each target T_J in order to arrive at the correct final match count.

Thus, for this embodiment, a match of the whole query tree against each nonrooted subtree list for T_X, T_Y, T_Z is made. The sum of the non rooted match counts is stored. Examining FIGS. 10a and 10b, in the example above, provides 1 match each from T_Y, T_Z

Following the non rooted matching, split the query tree into subtrees, one for each direct child of root. Children of query root are not combined as above, where target children of root are combined, since each target subtree may match against any single child of root in the query tree. For this embodiment, although the claimed subject matter is not limited in scope in this respect, it may be useful to construct and populate a grid of rooted subtree matches.

As described previously, all possible combinations of possible matches of rooted target partial subtrees against each child of root in the query tree is calculated. For a rooted match to occur, query children of root (row headings in table one below) are matched at least once against one or more of the rooted partial subtree lists maintained for each target subtree (column headings in table one below). The entry in the grid at a given row and column is the number of times that a single query tree child of root matches against a list of rooted partial subtrees for a single target subtree. In the example below, Table One, the entry values are taken from the example of FIGS. 10a and 10b.

	TABLE ONE
	TX	TY	TZ
	PA	1	1	0
	PB	1	0	1

For this embodiment, we also define P_J as a member of the set of all query subtrees (P_A, P_B in this example). For this embodiment, although the claimed subject matter is not limited in scope in this respect, the table may be stored in memory as a two dimensional array and may be traversed in column order, for example, recursively. A single combination may then comprise a single cell. For combinations, multiply all P_J together to obtain the match count. Of course, zero values may be omitted from the calculation for convenience. From the example table above, here we have the following matches:P_AT_X*P_BT_X=1 matchP_AT_X*P_BT_Z=1 matchP_AT_Y*P_BT_X=1 matchP_AT_Y*P_BT_Z=1 matchThe total match count, for this embodiment, therefore, is the sum of rooted and nonrooted match counts.

In accordance with the claimed subject matter, therefore, any tree, regardless of whether it is binary edge labeled, binary node labeled, non-binary, a feature tree, or otherwise, may be manipulated and/or operated upon in a manner similar to the approach of the previously described embodiments. Typically, different association embodiments shall be employed, depending at least in part, for example, upon the particular type of tree and/or string, as described, for example in the previously referenced U.S. provisional patent application 60/543,371. For example, as described in the previously referenced US provisional patent application, a node labeled tree in which the nodes are labeled with natural numerals or data values may be converted to a binary edge labeled tree. Furthermore, this may be accomplished with approximately the same amount of storage. For example, for this particular embodiment, this may involve substantially the same amount of node and/or edge data label values. However, for convenience, without intending to limit the scope of the claimed subject matter in any way, here, operations and/or manipulations and the like have been described primarily in the context of BELTs.

In another embodiment, however, a particular tree may include null types or, more particularly, some node values denoted by the empty set. This is illustrated, for example, by the tree in FIG. 7, although, of course, this is simply one example. Likewise, this example is an example of a binary node labeled tree with nulls, although, the claimed subject matter is not limited in scope in this respect. An advantage of employing null types includes the ability to address a broader array of hierarchical data sets. For example, without loss of generality and not intending to limit the scope of the claimed subject matter in any way, a null type permits representing in a database or a relational database, as two examples, situations where a particular attribute does not exist. As may be appreciated, this is different from a situation, for example, where a particular attribute may take on a numeral value of zero. Again, as described in the previously referenced U.S. provisional patent application 60/543,371, a tree with nulls, as described above, may be converted to a tree without nulls; however, the claimed subject matter is not limited in scope in this respect, of course. Thus, it may be desirable to be able to address both situations when representing, operating upon, manipulating and/or searching for patterns regarding hierarchical sets of data.

Likewise, in an alternative embodiment, a node labeled tree, for example, may comprise fixed length tuples of numerals. For such an embodiment, such multiple numerals may be combined into a single numeral, such as by employing Cantor pairing operations, for example. See, for example, Logical Number Theory, An Introduction, by Craig Smorynski, pp, 14-23, available from Springer-Verlag, 1991. This approach should produce a tree to which the previously described embodiments may then be applied. Furthermore, for one embodiment, a tree in which nodes are labeled with numerals or numerical data, rather than binary data, may be converted to a binary edge labeled tree and/or binary node labeled tree, and, for another embodiment, a tree in which edges are labeled with numerals or numerical data, rather than binary data, may be converted to a binary edge labeled tree and/or binary node labeled tree. See previously referenced U.S. provisional patent application Ser. No. 60/543,371.

Furthermore, a tree in which both the nodes and the edges are labeled may be referred to in this context as a feature tree and may be converted to a binary edge labeled tree and/or binary node labeled tree. For example, without intending to limit the scope of the claimed subject matter, in one approach, a feature tree may be converted by converting any labeled node with its labeled outgoing edge to an ordered pair of labels for the particular node. Using the embodiment described, for example in the previously referenced US provisional patent application, this tree may then be converted to a binary edge labeled tree.

In yet another embodiment, for trees in which data labels do not comprise simply natural numerals, such as, as one example, trees that include negative numerals, such data labels may be converted to an ordered pair of numerals. For example, the first numeral may represent a data type. Examples include a data type such as negative, dollars, etc. As described above, such trees may also be converted to binary edge labeled trees, such as by applying the embodiment of the previously referenced US provisional patent application, for example. However, again, this is provided for purposes of explanation and illustration. The claimed subject matter is not limited in scope to employing the approach of the previously referenced provisional patent application.

It will, of course, be understood that, although particular embodiments have just been described, the claimed subject matter is not limited in scope to a particular embodiment or implementation. For example, one embodiment may be in hardware, such as implemented to operate on a device or combination of devices, for example, whereas another embodiment may be in software. Likewise, an embodiment may be implemented in firmware, or as any combination of hardware, software, and/or firmware, for example. Likewise, although the claimed subject matter is not limited in scope in this respect, one embodiment may comprise one or more articles, such as a storage medium or storage media. This storage media, such as, one or more CD-ROMs and/or disks, for example, may have stored thereon instructions, that when executed by a system, such as a computer system, computing platform, or other system for example, may result in an embodiment of a method in accordance with the claimed subject matter being executed, such as one of the embodiments previously described, for example. As one potential example, a computing platform may include one or more processing units or processors, one or more input/output devices, such as a display, a keyboard and/or a mouse, and/or one or more memories, such as static random access memory, dynamic random access memory, flash memory, and/or a hard drive. For example, a display may be employed to display one or more queries, such as those that may be interrelated, and or one or more tree expressions, although, again, the claimed subject matter is not limited in scope to this example.

In the preceding description, various aspects of the claimed subject matter have been described. For purposes of explanation, specific numbers, systems and/or configurations were set forth to provide a thorough understanding of the claimed subject matter. However, it should be apparent to one skilled in the art having the benefit of this disclosure that the claimed subject matter may be practiced without the specific details. In other instances, well-known features were omitted and/or simplified so as not to obscure the claimed subject matter. While certain features have been illustrated and/or described herein, many modifications, substitutions, changes and/or equivalents will now occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such modifications and/or changes as fall within the true spirit of the claimed subject matter.

Claims

1. A method of performing tree matching comprising: executing instructions on a processor to:subdivide a target tree and a query tree;assign an individual and unique subdivided target tree numeral to at least one of subdivided target tree structures and assign an individual and unique subdivided query tree numeral to at least one of subdivided query tree structures, based at least in part on an association between trees and numerals, wherein the association between trees and numerals comprises to assign a list of individual and unique numerals to an associated list of individual and unique tree structures; andmatch the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral;wherein the executing instructions on a processor to subdivide a target tree and a query tree comprises to subdivide the trees so that the size of the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral does not exceed a numeral expressible within a platform native integer of fixed size for the processor.

2. The method of claim 1, wherein the numeral expressible within the platform native integer of fixed size corresponds to a size of data registers in a computing device comprising the processor.

3. The method of claim 1, wherein the size of the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral in a number of bits does not exceed a number of bits in the numeral expressible within the platform native integer of fixed size for the processor.

4. The method of claim 1, and further comprising executing instructions on the processor to: match another individual and unique subdivided query tree numeral with respect to another individual and unique subdivided target tree numeral.

5. The method of claim 1, wherein the trees comprising binary edge labeled trees (BELTs).

6. The method of claim 1, wherein the query tree comprising a partial subtree.

7. The method of claim 6, wherein the partial subtree comprising a rooted partial subtree.

8. The method of claim 6, wherein the partial subtree comprising a non-rooted partial subtree.

9. The method of claim 1, wherein the query tree comprising a full subtree.

10. The method of claim 1, wherein the executing instructions on the processor to match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral comprises to projection match the subdivided query tree numeral with respect to the subdivided target tree numeral.

11. The method of claim 1, wherein the executing instructions on the processor to match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral comprises to weighted match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral.

12. The method of claim 1, wherein the executing instructions on the processor to match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral comprises to partially match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral.

13. The method of claim 1, wherein the executing instructions on the processor to match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral comprises one of a closest match or a most remote match.

14. An article comprising: a non-transitory storage medium having stored thereon computer-readable instructions;a processor able to execute the computer-readable instructions, the executable computer-readable instructions to:subdivide a target tree and a query tree;assign an individual and unique subdivided target tree numeral to at least one of subdivided target tree structures and assign an individual and unique subdivided query tree numeral to at least one of subdivided query structures, respectively, with the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral based at least in part on an association between trees and numerals, wherein the association between trees and numerals comprises to assign a list of individual and unique numerals to an associated list of individual and unique tree structures; andmatch the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral;wherein the instructions to subdivide the target tree and the query tree comprises to subdivide the trees so that the size of the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral does not exceed a numeral expressible within a platform native integer of fixed size for the processor.

15. The article of claim 14, wherein the numeral expressible within the native integer of fixed size corresponds to size of data registers in a computing device comprising the processor.

16. The article of claim 14, wherein the size of the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral in a number of bits does not exceed a number of bits of the numeral expressible within the platform native integer of fixed size for the processor.

17. The article of claim 14, wherein the instructions further being executable by the processor to match another individual and unique subdivided query tree numeral with respect to another individual and unique subdivided target tree numeral.

18. The article of claim 14, wherein the trees to comprise binary edge labeled trees (BELTs).

19. The article of claim 14, wherein the query tree to comprise a partial subtree.

20. The article of claim 19, wherein the partial subtree to comprise a rooted partial subtree.

21. The article of claim 19, wherein the partial subtree to comprise a non-rooted partial subtree.

22. The article of claim 14, wherein the query tree to comprise a full subtree.

23. An apparatus comprising: a computing device, the computing device to subdivide a target tree and a query tree, to assign an individual and unique subdivided target tree numeral to at least one of subdivided target tree structures and an individual and unique subdivided query tree numeral to at least one of subdivided query tree structures, respectively, based at least in part on an association between trees and numerals, wherein the association between trees and numerals comprises to assign a list of individual and unique numerals to an associated list of individual and unique tree structures; and to match the individual and unique subdivided query tree numeral with respect to the individual and unique subdivided target tree numeral;wherein the computing device to subdivide the trees comprises to subdivide the trees so that a size of the individual and unique subdivided query tree numeral and the individual and unique subdivided target tree numeral do not exceed a numeral expressible within a platform native integer of fixed size for a processor.

24. The apparatus of claim 23, wherein the numeral expressible within the platform native integer of fixed size to correspond to a size of data registers in the computing device to comprise the processor.

25. The apparatus of claim 23, wherein the size of the individual and unique subdivided target tree numeral and the individual and unique subdivided query tree numeral in a number of bits does not exceed a number of bits for the numeral expressible for the platform native integer of fixed size for the computing device.

26. The apparatus of claim 23, wherein the computing device is further to match another individual and unique subdivided query tree numeral with respect to another individual and unique subdivided target tree numeral.

27. The apparatus of claim 23, wherein the trees to comprise binary edge labeled trees (BELTs).

28. The apparatus of claim 23, wherein the query tree to comprise a partial subtree.

29. The apparatus of claim 28, wherein the partial subtree to comprise a rooted partial subtree.

30. The apparatus of claim 28, wherein the partial subtree to comprise a non-rooted partial subtree.

31. The apparatus of claim 23, wherein the query tree to comprise a full subtree.