[{"data":1,"prerenderedAt":427},["ShallowReactive",2],{"project-6719582":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":46,"post_text":47,"what":48,"said":90,"maker_in_thread":195,"shape":197,"grid":200,"attempts":389,"same_week":390,"maker_shows":45,"maker_hidden":419,"reader_version":420,"card":421},"6719582","thisplusthat","A search engine that lets you 'add' words as vectors","title","Show HN: A search engine that lets you 'add' words as vectors","http:\u002F\u002Finsightdatascience.com\u002Fblog\u002Fthisplusthat_a_search_engine_that_lets_you_add_words_as_vectors.html","insightdatascience.com","jakek",1384280111,2013,148,64,133,1739,8,"notable","ai","word vector search","web_app",[24,25,26,27],"word2vec","word vectors","semantic search","word embeddings","curious",-0.034,{"positive":31,"neutral":32,"mixed":33,"critical":34},11,31,1,14,57,{"performance":37,"design":40,"idea":42,"execution":44},{"positive":38,"critical":39},5,10,{"positive":41,"critical":41},0,{"positive":43,"critical":33},4,{"positive":41,"critical":45},3,[33,45,45,33,45,33,33,38,45,45,45,38,45,33,45,38,45,45,45,38,41,33,41,45,33,45,33,33,33,38,33,38,45,38,45,45,45,38,38,45],null,{"problem":49,"audience":47,"pricing":47,"motivation":47,"time_to_build":47,"stack":61,"openness":87,"stage":88,"team_size":89},{"value":50,"source":51,"evidence":52},"Finds words related by vector arithmetic, like king - man + woman","commenter",[53,57],{"id":54,"by":55,"text":56},"6722478","est","Sounds like this paper from Google http:\u002F\u002Fwww.technologyreview.com\u002Fview\u002F519581 For example, the operation ‘king’ – ‘man’ + ‘woman’ results in a vector that is similar to ‘queen’.",{"id":58,"by":59,"text":60},"6720007","juxtaposicion","Hi, creator here (Chris Moody). Great question. The underlying algorithm, word2vec, (https:\u002F\u002Fcode.google.com\u002Fp\u002Fword2vec\u002F) isn't built for streaming data which means that at the…",[62,68,75,81],{"value":24,"source":51,"evidence":63},[64,65],{"id":58,"by":59,"text":60},{"id":66,"by":59,"text":67},"6721252","Hey doctoboggan, awesome questions! >> What is the dimensionality of each word vector and what does a words position in this space \"mean\"? What is this dimensionality determined…",{"value":69,"source":51,"evidence":70},"Python",[71],{"id":72,"by":73,"text":74},"6720666","doctoboggan","Hey juxtaposicion, fascinating work. I have many questions so I am just going to shoot them rapid fire. What is the dimensionality of each word vector and what does a words…",{"value":76,"source":51,"evidence":77},"Cython",[78],{"id":79,"by":59,"text":80},"6720359","Thanks! The number crunching is indeed very intense. The full body of vectors has a million rows and thousand columns, which fills up all of the 10GB of available memory. When…",{"value":82,"source":51,"evidence":83},"NumPy",[84],{"id":85,"by":59,"text":86},"6722430","Hey SandB0x, thanks for the advice! I actually started off by using numpy.dot, which is precisely what's needed. The problem is that I need it to go even faster (currently takes…","open_source","beta","unknown",{"praise":91,"objections":119,"requests":152,"questions":165,"alternatives":179},[92,107],{"value":93,"source":51,"evidence":94},"Interesting, fun results with some queries",[95,99,103],{"id":96,"by":97,"text":98},"6720319","grishma","Interesting. Currently it generates garbage for lot of queries but, some stuff is kinda fun. Forrest Gump - comedy + romance gives pulp fiction (!), as good as it gets (match)…",{"id":100,"by":101,"text":102},"6721704","nemo1618","Lisp + JVM = Clojure I'm sold. This is really cool! (Though it's worth noting that a Google search with the same terms returns the exact same result...)",{"id":104,"by":105,"text":106},"6724339","epaga","Pretty impressive for my first try. iPad - cool -> Windows Phone",{"value":108,"source":51,"evidence":109},"Concept is cool and relevant to related work",[110,111,115],{"id":100,"by":101,"text":102},{"id":112,"by":113,"text":114},"6723119","somberi","Fantastic work and is relevant to something we are working on in this space. Thanks. On a lighter note I tried \"sarah palin + sexy\" and I got John Mccain, Hillary Clinton and…",{"id":116,"by":117,"text":118},"6720256","emehrkay","Interesting. I've played around with words as vectors (with values) and the cosign similarity algorithm (http:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCosine_similarity). This is very cool stuff.…",[120,135,141],{"value":121,"source":51,"evidence":122},"Results are often nonsensical or uninspiring",[123,127,131],{"id":124,"by":125,"text":126},"6720200","gjm11","Hmm. It's a lovely idea but I find the results uninspiring. (Which isn't surprising; it's a very difficult problem. But I was hoping to be amazed.) Here are the examples I tried…",{"id":128,"by":129,"text":130},"6720147","leephillips","I was really excited by this writeup, so I tried it. Four test queries returned nothing that seemed useful or even relevant: fluid dynamics + electromagnetism : expected…",{"id":132,"by":133,"text":134},"6720369","tveita","I had the same experience. I haven't really gotten any convincing results except for the example queries verbatim. Some queries I tried: medicine - science + superstition…",{"value":136,"source":51,"evidence":137},"Limited training data hurts accuracy",[138],{"id":139,"by":59,"text":140},"6720243","It's not perfect; the real limiting factor is the volume of text. For the research paper behind word2vec, to get accurate associations for common words (King, man, woman, etc.)…",{"value":142,"source":51,"evidence":143},"Server crashes on non-ASCII input",[144,148],{"id":145,"by":146,"text":147},"6721785","cocoflunchy","Bug report: using some non-ascii characters crashes the server (for example é or É).",{"id":149,"by":150,"text":151},"6723927","jboynyc","This is neat, and I found a few queries that added interesting results. However, I tried ``` Slavoj Žižek - Jacques Lacan - Hegel ``` which yielded an internal server error,…",[153,160],{"value":154,"source":51,"evidence":155},"Access to the raw vector database",[156,159],{"id":157,"by":73,"text":158},"6738031","I'd love access to the raw vector database if you feel like sending it my way.",{"id":66,"by":59,"text":67},{"value":161,"source":51,"evidence":162},"Faster nearest-neighbour search",[163,164],{"id":66,"by":59,"text":67},{"id":79,"by":59,"text":80},[166,171],{"value":167,"source":51,"evidence":168},"How the vectors and dimensions work",[169,170],{"id":72,"by":73,"text":74},{"id":66,"by":59,"text":67},{"value":172,"source":51,"evidence":173},"How to handle changing content",[174,178],{"id":175,"by":176,"text":177},"6719906","donretag","Interesting concept, but how will it work with more dynamic content? You can train the model on a fairly static corpus such as Wikipedia, but what if you content changes with a…",{"id":58,"by":59,"text":60},[180,184],{"value":181,"source":51,"evidence":182},"Google search",[183],{"id":100,"by":101,"text":102},{"value":185,"source":51,"evidence":186},"Latent semantic indexing",[187,191],{"id":188,"by":189,"text":190},"6720059","toolslive","Does this relate to Latent Semantic Indexing? http:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLatent_semantic_indexing",{"id":192,"by":193,"text":194},"6720524","dhammack","Kind of. LSI is a dimensionality-reduction method, which means it takes really sparse \"bag of words\" vectors and compresses them in a way which approximates the original…",{"comments":41,"answered":41,"key":196},[],{"commenters":198,"max_depth":43,"maker_replies":41,"first_comment_mins":18,"lifespan_hours":199},52,95,[201,203,206,210,214,217,220,222,225,227,229,230,232,235,236,238,241,244,246,249,252,255,258,261,262,265,266,269,273,275,276,279,283,284,288,291,292,296,299,303,306,309,312,317,319,321,323,327,331,335,338,342,345,347,351,355,359,363,367,371,375,379,383,385],[124,41,33,202,125,41],"project\u002Fcritical\u002Fperformance",[204,41,45,205,59,41],"6720309","maker\u002Fneutral\u002Fperformance",[207,41,45,208,209,41],"6721868","project\u002Fneutral\u002Fdesign","riffraff",[211,41,45,212,213,41],"6722677","project\u002Fneutral\u002Fidea","pyduan",[215,41,33,202,216,41],"6723115","minikomi",[218,41,33,202,219,41],"6723196","mjfl",[100,33,38,221,101,41],"project\u002Fpositive\u002Fperformance",[223,33,33,202,224,41],"6745572","ccanassa",[128,226,33,202,129,41],2,[139,226,45,228,59,41],"maker\u002Fneutral\u002Fexecution",[132,226,33,202,133,41],[72,45,38,231,73,41],"maker\u002Fpositive\u002Fidea",[233,45,41,234,193,41],"6720858","other\u002Fneutral\u002Fnone",[66,45,45,228,59,41],[237,45,41,234,193,41],"6721354",[239,45,45,205,240,41],"6721936","murbard2",[242,45,45,205,243,41],"6722581","robseed",[157,45,38,245,73,41],"maker\u002Fpositive\u002Fnone",[247,43,45,248,59,41],"6719677","project\u002Fneutral\u002Fperformance",[250,43,45,248,251,41],"6719879","benmanns",[253,43,45,248,254,41],"6722575","joshschreuder",[256,43,33,202,257,41],"6724393","zamalek",[259,38,45,205,260,41],"6722078","SandB0x",[85,38,45,205,59,41],[116,263,38,264,117,41],6,"project\u002Fpositive\u002Fidea",[79,263,45,205,59,41],[267,263,45,205,268,41],"6720747","sillysaurus2",[270,263,45,271,272,41],"6721819","maker\u002Fneutral\u002Fnone","gregors",[175,274,45,212,176,41],7,[58,274,45,228,59,41],[277,18,38,221,278,41],"6720012","logn",[280,281,45,248,282,41],"6719786",9,"axblount",[54,39,41,234,55,41],[285,31,33,286,287,41],"6724026","project\u002Fcritical\u002Fexecution","jeorgun",[188,289,41,290,189,41],12,"other\u002Fneutral\u002Fidea",[192,289,41,290,193,41],[293,294,45,212,295,41],"6724156",13,"CurtMonash",[297,34,45,298,193,41],"6720566","maker\u002Fneutral\u002Fdesign",[300,301,33,202,302,41],"6720377",15,"Danieru",[304,301,41,305,117,41],"6720441","commenter\u002Fneutral\u002Fperformance",[307,301,45,248,308,41],"6723786","interpol_p",[96,310,45,311,97,41],16,"project\u002Fmixed\u002Fperformance",[313,314,33,315,316,41],"6723657",17,"project\u002Fcritical\u002Fidea","yetanotherphd",[149,318,33,286,150,41],18,[145,320,33,286,146,41],19,[322,320,45,228,59,41],"6722565",[324,325,45,248,326,41],"6722719",20,"zhemao",[328,329,38,264,330,41],"6721674",21,"breck",[332,333,33,202,334,41],"6720104",22,"akennberg",[336,333,41,305,337,41],"6723633","omaranto",[339,340,38,221,341,41],"6721227",23,"whistlerbrk",[343,340,45,248,344,41],"6725066","fredsanford",[112,346,38,264,113,41],24,[348,349,45,248,350,41],"6720282",25,"pit",[352,353,45,248,354,41],"6724298",26,"bocanaut",[356,357,45,248,358,41],"6724354",27,"corobo",[360,361,45,212,362,41],"6720912",28,"ppymou",[364,365,45,248,366,41],"6721325",29,"SergeyHack",[368,369,45,212,370,41],"6725275",30,"Lucy_karpova",[372,32,38,373,374,41],"6721161","project\u002Fpositive\u002Fnone","iLoch",[376,377,38,221,378,41],"6722139",32,"iamchmod",[380,381,45,248,382,41],"6721557",33,"elwell",[104,384,38,221,105,41],34,[386,387,33,202,388,41],"6724036",35,"dlsym",[],[391,396,401,406,410,415],{"id":392,"name":393,"points":394,"category":395},"6691316","StackDock",387,"infrastructure",{"id":397,"name":398,"points":399,"category":400},"6721474","giveit100.com",147,"productivity",{"id":402,"name":403,"points":404,"category":405},"6696654","Rather",98,"consumer_social",{"id":407,"name":408,"points":199,"category":409},"6689214","Shippo","web_saas",{"id":411,"name":412,"points":413,"category":414},"6727586","DataNitro",82,"data",{"id":416,"name":417,"points":418,"category":400},"6725995","ReRead",55,false,"v2",{"id":422,"name":5,"line":6,"ai":419,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":423,"top":18,"kw":424,"maker":11,"domain":10,"tier":19,"attempts":41,"strip":425,"url":9,"mix":426,"praise":93,"objection":121,"sub":21},6719582,-0.03,[24,25,26,27],"1331311533353135333501031311151535333553",[31,32,33,34],1794036511899]