I am fairly new to DS and Algorithms and recently at a job interview I was asked a question on performance tuning along with code. We have a Data Structure which contains multi-billion entries and we need to search a particular word in that data structure. So which Java feature/library can we use to do the searching in the quickest time possible ?
On the spot I could not think of exact answer so I wrote that:
We can store the values in a map and search words in the map (but got stuck how to decide key-value pair in the map).
How can I understand the exact answer to this question and what can be the optimal solution(s) ?
After reading the question and getting clarification in the comments, I think what has become apparent to me is that: you needed to ask follow-up questions.
I'll try to break it down and provide comments that I hope will be helpful, because I also know what it's like to be "in the moment" and how nerves can stab you in the back when you least need them to.
We have a Data Structure which contains multi-billion entries and we need to search a particular word in that data structure.
I think a good follow-up question here would've been:
Q: What specific data structure is being used to contain all this data?
I would press until they give me an actual name and explain why it is not possible to name a Java algorithm/library. For all you know, the data structure could've been String[], a Set<String>, or even a fancy name for a file on disk (if they're trying to throw you off). They could've also clarified and said the DS was not relevant and that you could pick whichever DS you thought was best.
The wording also implies that they implemented the structure and that it's already populated in a system with, presumably, enough memory to hold all of it. Asking to confirm that this is really the case could've given you helpful information.
For example: "Based on the wording, it seems this mystery data structure is already implemented and fully populated in memory in a system with enough memory to hold it. Can you confirm my understanding here is correct? If not, could you clarify further?"
Given the suggested wording, and the fact that we don't have additional clarifications to go from, I will assume, for the purposes of this answer, that my suppositions are indeed correct.
Note that if you had been asked to design the data structure to hold all of this info, you would've had to ask very different questions, take memory constraints into account, and perhaps even ask about character sets/encodings (e.g. ASCII vs multi-byte Unicode).
Also, if you had been asked to design the search algorithm, then knowing the DS is a pre-requisite, and not knowing this could've made the task impossible. For example, the binary search algorithm implementation will look very different if you're working on an array vs a binary search tree, even though both would offer O(lg n) time complexity.
So which java feature/library can we use to do the searching in the quickest time possible?
Consistent with the 1st part, this question only asks what pre-existing/built-in Java code you would choose to perform the search for you. The "quickest time possible" here should make you think about solutions that are in O(1), i.e. are constant time. However, the data structure may open/close doors for you.
Some search algorithms in Java work on generics and others work on other types like arrays. Some algorithms work on Maps while others work on Lists, Sets, and so on. The follow-up question from the first part could've helped in answering this question.
That said, even if you knew the DS, but couldn't think of a specific method name or such at the time, I also think it should be considered reasonable to mention the interface or at least a relevant package and say that further details can be checked on the the Java documentation if you're pressed for more specificity, given that's what it's there for in the first place.
We can store the values in a map and search words in the map (but got stuck how to decide key-value pair in the map).
Given the wording, my interpretation of their question was not "which data structure would you use?", but rather, "which pre-existing search algorithm would you choose?". It seems to me like it was them who needed to answer the question regarding DS.
That said, if you had indeed been asked "which data structure would you use?", then a Map would've still worked against you, since you didn't really need to map a key to a value. You only needed to store a value (i.e. the words). Therefore, a Set, specifically a HashSet, would've been a better candidate, since it also avoids duplicates and should consume less memory in the process because it stores singular values, rather than key/value pairs.
Of course, that's still under the assumption(s) I made earlier. If memory constraints are said to be an issue, then scaling horizontally to multiple servers and so on would've likely been necessary.
How can I understand the exact answer to this question and what can be the optimal solution(s)?
It is probably the case that they wanted to see if you would follow up with questions, given the lack of information they gave you.
There are a couple data structures that allow for efficient searching, assuming that memory requirements aren't an issue and the data structure is already populated.
Regarding time complexity, Set#contains and Map#containsKey are both O(1), assuming that the hash function isn't expensive and that there aren't many collisions.
Because the data structure stores words (assuming you're referring to Strings), then it could also be relatively efficient to use a trie (radix tree, prefix tree, etc.), which would allow you to search by character (which I believe would be O(log n)). If the hash function is expensive or there are many collisions, this could be a good alternative!
The answer that you gave to the interviewer should suffice since hashing is an effective searching method, even for billions of entries.
You did not mention whether the entries are words or documents (multiple words). In both cases a search index could be suitable.
Search indexes extract words from the billion document entries and manage a map of these words to the documents they are used in. Frameworks like Lucene (e.g. as part of SOLR or ElasticSearch) manage memory and persistence for you.
If it were only multiple of thousands of entries, a simple HashMap would be sufficient because there is no need for memory management then. If all of the billion entries are single words, a database could be a slightly better choice.
The hashmap solution is reasonable as stated by others but there are doubts with respect to scalability.
Here is a possible solution for the problem as discussed in the below post
Sub-string match If your entry blob is a single sting or word (without any white space) and you need to search arbitrary sub-string within it. In such cases you need to parse every entry to find best possible entries that matches. One uses algorithms like Boyer Moor algorithm. See this and this for details. This is also equivalent to grep - because grep uses similar stuff inside
Indexed search. Here you are assuming that entry contains set of words and search is limited to fixed word lengths. In this case, entries are indexed over all the possible occurrences of words. This is often called "Full Text search". There are number of algorithms to do this and number of open source projects that can be used directly. Many of them, also support wild card search, approximate search etc. as below :
a. Apache Lucene : http://lucene.apache.org/java/docs/index.html
b. OpenFTS : http://openfts.sourceforge.net/
c. Sphinx http://sphinxsearch.com/
Most likely if you need "fixed words" as queries, the approach two will be very fast and effective
Reference - https://softwareengineering.stackexchange.com/questions/118759/how-to-quickly-search-through-a-very-large-list-of-strings-records-on-a-databa
Multi-billion entries lie at the edge of what might conceivably be stored in main memory (for instance, storing 10 billion entries at 100 bytes per entry will take 1000 GB main memory).
While storing the data in main memory offers a very high throughput (thousands to millions of requests per second), you'd likely need special hardware (typical blade servers only offers 16 GB, but there are commodity servers that permit installation of up to 3000 GB of main memory). Also, keeping this much data in the Java Heap will likely cause garbage collector pauses of seconds or minutes unless special care is taken.
Therefore, unless the structure of your data admits a very compact representation in main memory (say, you only need membership checking among ints, which is possible with a 512 MB Bitset), you'll not want to store it in main memory, but on disk.
Therefore, you'll need persistence. Any relational or NoSQL database permits efficient searching by key and can handle such amounts of data with ease. To talk to a relational database, use JPA or JDBC. To talk to a non-relational database, you can use their proprietary Java API or an abstraction layer such as Spring Data.
You could also implement persistence from scratch if you wanted to (i.e. the interviewer asks for that). A data structure optimized for efficient lookup in external memory is the B-Tree, that's what many databases use internally :-)
Related
I decided to revise Java collection framework, so I started with internal implementation. One question came on my mind, which I can't solve. Hope someone can make a clear explanation on following.
ArrayList uses linear or binary search (both have pros/cons), but we can do anything with them! My question is why do all 'hashing' classes (like HashMap f.e.) use hashing principle? Couldn't they settle with linear or binary search for example? Why just not store Key/Value pair inside array? And the opposite, why isn't (for example ArrayList stored in hashTable)?
The intention of the collections framework is that the programmer will choose the data structure appropriate to the use case. Depending on what you're using it for, different data structures are appropriate.
Hashing classes use the hashing principle, as you put it, because if you choose them, then that's what you want to use. (Hashing is generally the best choice for simple, straightforward lookups.) A screwdriver uses the screwing principle because if you pick up a screwdriver, you want to screw something in; if you had a nail you needed to put in, you would have picked up the hammer instead.
But if you're not going to be performing lookups, or if linear search is good enough for you, then an ArrayList is what you want. It's not worth adding a hash table to a collection that's never going to use it, and it costs CPU and memory to do things you aren't going to need.
I had a large hash of values (about 1,500). The nature of the code was that once the hashmap was loaded it would never be altered. The hashmap was accessed many times per web page, and I had wondered if it could be sped up for faster page loading.
One day I had some time, so I did a series of time tests (using the nano time function). I then reworked the hashmap use over to an array. Not an ArrayList, but an actual array[]. I stored the index with the key class used to get the hash value.
There was a difference, that the array lookup was faster. I calculated that over a days worth of activity I would have saved almost a full second!
So yes, using an array is faster than using a hash, YMMV :-)
And I reverted my code back to using a hashmap, as it was easier to maintain...
To make it simple, my question is: how to hash a String (about 200 characters) as quickly as possible. Security is not important, but collisions ARE a big deal.
Note: After a quick investigation, it seems like MurmurHash3 might be the best choice. I am open to any comment saying otherwise tho'
First, I know that there are plenty of other similar question, but I couldn't find a convincing answer yet.
I have a list of objects, each containing a list of about 3k paragraphs which is saved to a database. Every X hours, those paragraph are regenerated and I need to find if any paragraphs has changed, and if so, push only those new paragraphs.
The quickest way I found to find the differences (knowing that most of the time the content will be identical) is to create a MerkleTree, save it to the DB, and iterate over the MerkleTree to find the differences, instead of comparing the paragraphs themselves.
This imply, in my case, that I will be creating ten thousands of hashes per second to compare with what is in the DB. Therefore, I need a very efficient way to create those hashes. I don't care about the security, I only need to ensure that the number of collision remains very very low.
What would be the best algorithm available in Java for that?
In my case, the main object is composed of Sections, which is composed of Languages, which is composed of Paragraph. The comparison strategy is:
1) If the object hash is identical, stop, otherwise go to 2)
2) Loop on all Section, keep only the Section with a different hash
3) Loop on all Languages of those Sections, keep only the language with a different hash
4) Loop on all the Paragraph of all those Languages, if the hash is different, then push the new content.
This amazing answer on Programmers Stack Exchange tells you all you need to know.
The short version is, use FNV-1a, aka the Fowler–Noll–Vo hash function, it has excellent performance, high randomness and low collisions.
Any further explanation I might shed on this question would be just be a copy and paste from that Programmers.SE answer, which incidentally is the second highest voted answer on the entire site.
Some other thoughts:
Ultimately, you have a pretty niche use case. Most people aren't dealing with 1 billion entry datasets regularly. As such, you may have to do your own benchmarking.
That said, having a high randomness suggests that the algorithm is likely to scale well for English hashes.
You haven't really talked about other issues; are you able to keep the entire data set in memory? What are your footprint requirements?
See also: Fastest Hash Algorithm for Text Data
In an interview recently, I was asked about how HashMap works in Java and I was able to explain it well and explain that in the worst case the HashMap may degenerate into a list due to chaining. I was asked to figure out a way to improve this performance but I was unable to do that during the interview. The interviewer asked me to look up "Trove".
I believe he was pointing to this page. I have read the description provided on that page but still can't figure out how it overcomes the limitations of the java.util.HashMap.
Even a hint would be appreciated. Thanks!!
The key phrase there is open addressing. Instead of hashing to an array of buckets, all the entries are in one big array. When you add an element, if the space for it is already in use you just move down the array to find a free space.
As long as the array is kept sufficiently bigger than the number of entries and the hash function is well distributed it's possible to keep average lookup times small. And by having one array you can get better performance - it's more cache friendly.
However it still has worst-case linear behaviour if (say) every key hashes to the same value, so it doesn't avoid that issue.
It seems to me from the Trove page that there are two main differences that improve performance.
The first is the use of open addressing (http://en.wikipedia.org/wiki/Hash_table#Open_addressing). This doesn't avoid the collision issue, but it does mean that there's no need to create "Entry" objects for every item that goes in the map.
The second important difference is being able to provide your own hash function, which differs from the one provided by the class of the keys. So you could provide a much faster hash function if it made sense to do so.
One advantage of Trove is that it avoids object creation, especially for primitives.
For big hashtables in an embedded java device this can be advantageous due fewer memory consumption.
The other advantage, I saw is the use of custom hash codes / functions without the need to override hashcode(). For a specific data set, and an expert in writing hash functions this can be an advantage.
I hope that this question is specific enough to be deemed fit for StackOverflow. I checked the FAQ and I think this qualifies, since it is specific and related to programming.
I'm implementing a complex data mining algorithm (FP-growth) in Java. Some of the initial phases of the algorithm require me to scan a large database and keep a running count of each item type found. This seems perfectly suited to a Hashbag interface. I found one in Apache Commons which seems to work for me.
So now, my HashBag is filled with [itemType, count] entries (pairs). Later on in the algorithm, I'm required to do a lot of list-like operations on these pairs. In some cases, I must sort the collection by itemType. In others, I must sort by count. This seems perfectly suited to a List interface.
I'm left with the conclusion that I must convert my Hasbag to a List. Yet it feels dirty somehow, like a waste of space and time. Is there a smarter way to do this, or is it a common situation to have a programming problem where you must treat your collection differently at different times, and conversions are a necessary evil?
One alternative is to make my own interface which is truly a list, but allows "bag-style" adds. I'd have to keep the list sorted and perform binary searches with a custom comparator every time I wanted to add something. Building that collection would probably take longer than building a Hashbag, but I'd save on the conversion step at the end. Any thoughts as to which is preferable?
Thanks!
If you used Guava's Multiset instead of Apache's Bag -- roughly analogous, but in a different style -- you can do most of this without converting. Multiset.entrySet() returns a Set<Entry<E>>, with Entry<E> effectively representing a pair of an element and a count -- that sounds like it's probably the best way to address your need to operate on the element-count pairs, maybe? You can iterate over that like you'd iterate over a Map.entrySet().
You can use Multisets.copyHighestCountFirst(Multiset) to get a multiset reordered in highest-frequency-first order, and use TreeMultiset to order by the elements directly.
(Disclosure: I contribute to Guava.)
I assume you're using the Apache Commons Collections HashBag class. Have you considered using TreeBag instead? It implements the same Bag interface but efficiently keeps the data sorted according to a comparator you provide.
That said, when you need to change sort order, there isn't usually any better answer than to copy the collection to a new one with a different comparator.
Yet it feels dirty somehow, like a waste of space and time. Is there a smarter way to do this, or is it a common situation to have a programming problem where you must treat your collection differently at different times, and conversions are a necessary evil?
Sometimes it is necessary to convert between collection types. If it is necessary "dirty" or "inelegant" or "dumb" are not really relevant.
It can also be a mistake to over-think these things up front. The actual computational trade-offs are often difficult to grasp. For instance, if you changed the HashBag to a TreeBag, insertion goes from O(1) to O(logN) but you then avoid the overheads of sorting and copying. "Big Oh" analysis / thinking is not going to give you a clear answer. Indeed, the real performance is going to depend on the scaling factors, the values of N, the ratio of hits and misses in the bag and so on.
I would advise to try implementing things the obvious way, and see if it performs well enough ... and if not, profile it to see if the data structures are the main bottleneck. Then based on the profiling, and other measurements of the input datasets, figure out the best way to improve performance from your baseline implementation.
Answering my own question!
I did some experimenting with the different types of Multiset provided by the Guava libary mentioned above by Louis Wasserman. In my particular test case, I'm parsing a 1GB XML file (database of books and authors) and creating a very large Multiset (keeping a count of how many times each author shows up in the DB). Once I reach the end of the parsing, I need to get a new Multiset which only contains the authors who showed up more than x times, where x is some threshold value. I also want my final set to be sorted by author name.
Here are two of the different ways I tried it (among others):
1) collect the original counts in a TreeMultiset and then remove any which don't meet the threshold
2) collect the original counts in a HashMultiset, and then create a new TreeMultiset where I add each item from the hash set with a count the meets the threshold
The second way proved to be significantly faster (roughly 25%), despite the conversion and extra memory usage. Obviously a big part of this is that it is pretty inefficient to delete from binary trees.
So the clear conclusion here is that in this situation, conversion is a good move (unless you have memory constraints that won't allow it).
Thanks again for turning me onto the Guava library, Louis!
I just realized that in my 4+ years of Java programming (mostly desktop apps) I never used the binary search methods in the Arrays class for anything practical. Not even once. Some reasons I can think of:
100% of the time you can get away with linear search, maps or something else that isn't binary search.
The incoming data is almost never sorted, and making it sorted requires an extra sorting step.
So I wonder if it's just me, or do a lot of people never use binary search? And what are some good, practical usage examples of binary search?
On the desktop, you're probably just dealing with the user's data, which might not be all that big. If you are querying over very large datasets, shared by many users, then it can be a different matter. A lot of people don't necessarily deal with binary search directly, but anyone using a database is probably using it implicitly. If you use AppEngine, for example, datastore queries almost certainly use binary search.
I would say it boils down to this:
If we are going to do a binary search, we're going to have a key to search by. If we have a key, we're probably using a map instead of an array.
There's another very important thing to keep in mind:
Binary search is a clear-cut example of how thinking like a good programmer is very different than thinking like a normal person. It's one of those cognitive leaps that makes you really think about taking operations that are traditionally done (when done by humans) in order-n time and taking it down to order-lg-n time. And that makes it very, very useful even if it's never used in production code.
I hardly ever, if ever use a binary search.
But I would if:
I needed to search the same list multiple times
the list was long enough to have a performance problem (although I'm often guilty of micro-optimization)
However, I often use hash tables / dictionaries for fast lookups.
For production code on my day job, a Set or Map is always good enough so far.
For algorithmic problems that a I solve for fun, binary search is a very useful technique. For starters, if the set of elements never changes (i.e. you are never going to insert or delete elements in the set being queried) a Map/Set has no advantage over binary search - and a binary search over a simple array avoids a lot of the overhead associated with querying a more complex data structure. In many cases I have seen it to be actually faster than a HashMap.
Binary search is also a more general technique than just querying for membership in a set. Binary search can be performed on any monotone function to find a value for which the function satisfies a certain criteria. You can find a more detailed explanation here. But as I said, my line of work does not bring up enough computationally involved problems for this to be applicable.
Assume you have to search an element in a list.
You could use linear search, you’ll get O(n).
Alternatively, you could sort it by fastest algorithm (O(log n)*n), and binary search(O(log n)). You’ll get O((log n)*n + log n).
That means when searching large size of list, binary search is better. Also, it depends data structure of list. If list is a link based list, binary search is bad practice.